THE COLOSSAL HATS (PUKAO) OF MONUMENTAL 
STATUES: AN ANALYSIS OF PUKAO VARIABILITY ON 
RAPA NUI (EASTER ISLAND) 
 
 
 
 
 
 
 
 
by 
SEAN HIXON 
 
 
 
 
 
 
 
 
 
 
A THESIS 
 
 
 
 
 
Presented to the Department of Anthropology  
and the Robert D. Clark Honors College  
in partial fulfillment of the requirements for the degree of  
Bachelor of Science 
 
June, 2015 
 
An Abstract of the Thesis of 
Sean Hixon for the degree of Bachelor of Science 
in the Department of Anthropology to be taken June, 2015 
Title: The colossal hats (pukao) of monumental statues: an analysis of pukao on Rapa Nui (Easter 
Island) 
As part of monumental statue (moai) construction during the prehistory of Rapa Nui, 
islanders quarried masses of red scoria, carved them into hats (pukao ), and placed them 
atop statues measuring up to 10 meters tall. Despite overall great interest in moai and 
the improbable magnitude of pukao that were raised to reach their positions on the 
heads of statues, few studies have investigated pukao production, transport, and 
placement. This study seeks to analyze three-dimensional variability of pukao using 
15,000 photos of 50 pukao found near statues and 13 red scoria cylinders located in the 
quarry. These models are used to evaluate which surface features are stylistic with 
related temporal and spatial variability and which are functional and relate to 
construction and transport of these multi-ton objects. The functional detail has the 
potential to shed light on how prehistoric islanders designed pukao to be placed atop 
moai. To this end, additional three dimensional models of statue platforms (ahu) and 
moai are combined with the models of pukao to test the feasibility of the hypothesis that 
the pukao were placed atop moai using stone ramps. The hypothesis in which relatively 
small numbers of Rapanui placed pukao atop moai by rolling the former up ramps is 
supported by physical calculations and best explains the archaeological record. 
  
iii  
 
 
 
 
Acknowledgements 
I would like to thank Professors Terry Hunt and Carl Lipo for their help in data collection and 
for their ongoing guidance.  I must also thank the entire 2014 NSF-REU Geospatial Research and 
Mapping group for their support during our stay on Rapa Nui.  I am also very grateful for the financial 
support for this research that came from the University of Oregon Center for Teaching and Learning 
Undergraduate Research Fellowship.  I would also like to thank Professors Ben McMorran, Stephen 
Dueppen, and Scott Fitzpatrick, as well as Robert DiNapoli, John O’Connor, Damion Sailors, and Brian 
Lane for their guidance in considering various aspects of this project.  Finally, I must thank Nastassia 
Donoho and my parents, Mark Hixon and Ursula Bechert, for their editing and support. 
 
  
iv  
 
 
 
 
Table of Contents 
Introduction 1 
The Conventional Narrative 4 
Limited Historic Record of Pukao 6 
Speculation on Pukao Meaning 7 
Prior Thoughts on Pukao Transport 13 
Broad Significance of Research 22 
Approach 27 
Evolutionary Framework for Understanding Style and Function 28 
Applications to Pukao 30 
Methods 34 
Model Generations 34 
Recording Strategy 36 
Results 40 
Sample Description 40 
Issues of Preservation 42 
Human Factors 42 
The Elements 45 
Non-Human Biological Factors 50 
Note on Terminology 52 
Volume and Weight 53 
Base Indentations 56 
Physical Form 60 
Body 62 
Top Cylinder 67 
Surface Markings 71 
Side Striations 71 
Cupules 77 
  
v  
Puna Pau Scoria Bodies 81 
Other Sources of Variation 82 
Petroglyphs 82 
Slabs Missing 91 
Basins 93 
Englert Numbers 97 
Western Text Engravings 98 
Conclusion 99 
Implications of Pukao Dimensions for Transport 99 
Volume 99 
Base Indentation 99 
Physical Form 100 
Surface Variation 101 
The Modified Ramp Hypothesis 102 
Physics Derivation: Pushing through Levering 105 
Physics Derivation:  Pulling with Rope 111 
Implications of Transport Equations 113 
Physics Derivation. Ramp Volume Estimates 118 
Implications of Ramp Volume Estimates 121 
Modifications to Ramp Hypothesis 125 
Future Research 130 
Summary 134 
Appendix 136 
Model Generation, Filing, and Observation Strategies 136 
Additional Physics Derivation: Final Flipping of Pukao 138 
Bibliography 140 
 
  
vi  
 
List of Figures  
Figure 1  Map showing terrain, major archaeological features, and infrastructure of 
Rapa Nui. 1 
Figure 2.  Map showing distances between Rapa Nui and nearest landmasses. 3 
Figure 3.  Photograph of red scoria quarry at Puna Pau. 3 
Figure 4.  Photograph of red scoria pukao atop moai on an ahu at Anakena, located on 
the northern side of the island. 4 
Figure 5.  Walking of a replica moai at Kualoa Ranch, O’ahu. 6 
Figure 6.  Engraving based on a 1778 sketch by William Ellis from Hawai’i during 
James Cook’s Third Voyage. 9 
Figure 7.  A 1777 engraving based on a sketch by William Hodges from New Caledonia 
during James Cook’s Second Voyage. 10 
Figure 8.  Image from Anakena facing W and giving a view of a large scoria body 
located at a distance from the nearest ahu. 14 
Figure 9.  William Mulloy’s hypothesized method of prehistoric red scoria body 
movement on Rapa Nui. 15 
Figure 10.  Cartoon depiction of a possible rendition of the Ramp Hypothesis. 18 
Figure 11.  Depiction of the Tower Hypothesis, which involves a pile of stones built 
beneath the pukao. 20 
Figure 12.  Depiction of the Simultaneous Set Up Hypothesis in which the moai and 
pukao are erected simultaneously. 21 
Figure 13.  Depiction of Wooden Ramp Hypothesis in which pukao is slid up a wooden 
ramp to the top of a moai. 22 
Figure 14.  Writers such as Erich von Däniken, whether they intend to or not, often 
remove the agency of the prehistoric Rapanui in monumental statue production by 
invoking highly speculative tales of extraterrestrials. 24 
Figure 15.  A sample of wrenches illustrating stylistic modes that vary from wrench to 
wrench (e.g., composing metal) and functional modes that are shared by all (e.g. 
hexagonal notch). 29 
Figure 16.  Hypothetical changes in frequency of modes under selection versus modes 
under drift. 32 
Figure 17.  Illustration of the method behind structure for motion mapping. 35 
Figure 18.  One of 290 photos (left) used to generate a textured model of pukao 18 at 
Vaihu in Photoscan (right). 36 
Figure 19.  Textured model of pukao 18 colored in Meshlab according to the APSS 
filter in order to highlight surface details. 37 
  
vii  
Figure 20.  Map showing locations on Rapa Nui that yielded usable models of scoria 
bodies (green) and locations that include only pukao fragments that were not modelled 
(red). 40 
Figure 21.  Graph showing the number of models generated from each location (black) 
and the number of models at each location that were not modeled (grey). 41 
Figure 22.  Red scoria reused in Catholic cemetery in Hanga Roa and replica pukao at 
Tahai. 44 
Figure 23.  Diagram illustrating expansion of a clay mineral with a sheet structure. 46 
Figure 24.  Photograph of pukao in intertidal zone at Vaihu prior to salvage efforts 
during Norwegian Archaeology Expedition in 1955. 47 
Figure 25.  Photographic panorama of the front of the ahu showing fallen moai and 
pukao at Tongariki in 1914-15. 49 
Figure 26.  Photographs of restored ahu at Tongariki with a single moai with restored 
pukao (top) and nearby alignment of pukao following the 1960 tsunami. 49 
Figure 27.  Pukao 55 at Tongariki covered in lichen. 50 
Figure 28.  Undated photograph of sheep roaming Rapa Nui. 51 
Figure 29.  Chart illustrating the number of scoria bodies within each condition mode 
and representative samples from the extreme modes. 52 
Figure 30.  Photographs of the two relatively large cylindrical bodies of red scoria 
included in the pukao database and not located in Puna Pau that are not considered to be 
pukao. 53 
Figure 31.  Volume of pukao plotted as a function of distance from quarry. 54 
Figure 32.  Cumulative number of red scoria cylinders as a function of distance from 
quarry. 55 
Figure 33.  Photograph collected by Métraux of the pukao belonging to Paro at Te Pito 
Kura. 56 
Figure 34.  Chart illustrating the number of scoria bodies within each resting face mode, 
and representative samples of a few of the modes. 58 
Figure 35.  Sketches of base (left) and side (right) views of pukao and the variables 
considered. 59 
Figure 36.  Fragments from engravings based on sketches made by visitors to Rapa Nui 
during Cook’s 1774 visit and La Pérouse’s 1786 visit that highlight pukao. 61 
Figure 37.  Sector division of pukao-bearing ahu made for use in spatial comparisons.  
 62 
Figure 38.  Photograph of the noticeably conical pukao 65 at Anakena. 63 
Figure 39.  One-way analysis of variance and pairwise multiple comparison applied to 
measures of how oblong pukao body cross sections are between sectors. 65 
  
viii  
Figure 40.  One-way analysis of variance and pairwise multiple comparison applied to 
measures of how squat pukao bodies are between sectors. 66 
Figure 41.  Spatial comparison of pukao body and top cylinder mean dimensions by 
sector with shape aids for visualization. 667 
Figure 42.  Pukao 66 at Anakena exhibiting an inverted conical protuberance when 
viewed from the front. 68 
Figure 43.  One-way analysis of variance and pairwise multiple comparison applied to 
measures of how oblong top cylinder body cross sections are between sectors. 70 
Figure 44.  One-way analysis of variance and pairwise multiple comparison applied to 
measures of how squat top cylinders are between sectors. 71 
Figure 45.  Representative samples of a few of the modes within the dimension of side 
striations. 72 
Figure 46.  Representative samples of a few of the modes within the dimension of 
cupules. 77 
Figure 47.  A watercolor from Linton Palmer’s 1868 visit to the island showing a few of 
the Puna Pau scoria bodies. 83 
Figure 48.  Various tracings of petroglyphs on red scoria pukao, Puna Pau scoria bodies, 
and red scoria ahu lentils, organized by location. 84 
Figure 49.  Various pukao with isolated curved line petroglyphs. 86 
Figure 50.  Various scoria bodies with closed curved form petroglyphs. 88 
Figure 51.  Pukao 4 from Vinapu with isolated curved lines, a cupule, and circles at the 
lower left. 89 
Figure 52.  Drawing by P. Cristián Arévalo of pukao 41 at Tu’u Tahi with horizontally 
aligned side cupules. 90 
Figure 53.  The four pukao that include horizontally-aligned pairs of side cupules near 
their bases.  90 
Figure 54.  Four of the pukao that have large slabs missing from the upper body edges.  
 92 
Figure 55.  Scoria body 86 from Puna Pau in upper photo at right with concave hollow 
missing. 93 
Figure 56.  The two pukao located along the southern coast with basins dug. 94 
Figure 57.  Shallow excavation conducted by the Norwegian Archaeological Expedition 
in 1955 around the base of pukao 4. 95 
Figure 58.  Top views of three pukao with notable circular holes in the center of the 
upward face. 96 
Figure 59.  Two scoria bodies with clearly-marked Englert numbers. 98 
  
ix  
Figure 60.  Two scoria bodies at Puna Pau (77 at left and 80 at right) with engraved 
Western text. 98 
Figure 61.  The configuration of a parbuckle. 105 
Figure 62.  Forces involved in levering a pukao of radius R up a ramp, where  gives 
the angle of incline of the ramp. 106 
Figure 63.  Torques acting on the lever as it pivots around point P. 108 
Figure 64.  An analysis of the distance d and how it relates to . 109 
Figure 65. Forces involved in pulling a pukao of radius R up a ramp, where  gives the 
angle of incline of the ramp. 112 
Figure 66.  Equation giving the force required to move a pukao up a ramp by pushing 
with a lever (left, equation 16) or by pulling with a parbuckle (right, equation 18). 114 
Figure 67.  Applied force required to move pukao up ramps with variable angles of 
incline at four different ahu (1 = Te Pito Kura, 2 = Vaihu, 3 =  Akahanga, and 4 = Ura 
Uranga te Mahina). 115 
Figure 68.  Applied force required to pull pukao up ramps with variable angles of 
incline at four different numbered ahu (1 = Te Pito Kura, 2 = Vaihu, 3 =  Akahanga, 
and 4 = Ura Uranga te Mahina). 117 
Figure 69.  A ramp with an angle of incline of , height h, width w, and length a. 118 
Figure 70.  Equation for the volume of a ramp. 121 
Figure 71.  Ramp volume as a function of ramp angle of incline. 122 
Figure 72.  Scale sketch by Heyerdahl and Ferdon showing the extent of the ahu at Te 
Pito Kura. 123 
Figure 73.  Spread of stones at the ahu at Ura Uranga te Mahina measuring about 13 
meters in length. 125 
Figure 74.  Image illustrating the convergence of two flat surfaces (marked with black 
lines) on a fallen moai base just east of Tongariki. 126 
Figure 75.  Scale sketch showing an approximation of a reasonable ramp coming off of 
the front of a forward-leaning moai at the Vaihu ahu. 127 
Figure 76.  Sketch of side view of moai and pukao showing the vertical projection of 
the pukao center of mass (CM) when given a forward moai lean angle of phi. 129 
Figure 77.  Textured model of pukao 57 from Hekii showing the pronounced back edge 
of its base indentation (about 4 centimeters deep) in the upper right and the relatively 
indistinct front edge of its base indentation near the center. 130 
Figure 78.  Various dating methods and their effective ranges. 133 
  
x  
 
List of Tables  
Table 1.  List of various uses and reuses of red scoria on Rapa Nui. 13 
Table 2.  Five proposed pukao transport mechanisms with their supporters and basis 
through time. 16 
Table 3.  The pukao transport hypotheses listed in Table 2 and expectations for 
associated wear. 28 
Table 4.  Dimensions and modes of categorical variables. 38 
Table 5.  Summary of explanations as to why scoria bodies in given locations were not 
effectively modelled. 42 
Table 6.  Distances and ratios of distances between pukao features and centers of mass.  
 60 
Table 7.  Tapering in the diameters of pukao with the form of truncated cones. 63 
Table 8.  Tapering in the inverted conical protuberances of two pukao. 68 
Table 9.  Comparison of number of scoria bodies with all sides exposed and striations 
present within each sector relative to total number of scoria bodies with all sides 
exposed within each sector. 73 
Table 10.  Comparison of number of scoria bodies without all sides exposed and 
striations present within each sector relative to total number of scoria bodies without all 
sides exposed within each sector. 73 
Table 11.  Abundances of different types of side striations within each sector for scoria 
bodies with all sides exposed. 74 
Table 12.  Abundances of different types of side striations within each sector for scoria 
bodies of all resting faces. 74 
Table 13.  Results from Fisher tests of individual side striation mode abundances 
between sectors. 75 
Table 14.  Overall abundances of the three types of side striations when considering all 
scoria bodies and only those with fully exposed sides separately. 76 
Table 15.  Comparison of number of scoria bodies with cupules present within each 
sector relative to total number of scoria bodies within each sector. 78 
Table 16.  Abundances of different types of cupules within each sector for scoria bodies 
of all resting faces. 79 
Table 17.  Results from Fisher tests of individual cupule mode abundances between 
sectors. 79 
Table 18.  Observation matrix listing intersections of side cupule and condition modes.
 81 
Table 19.  Comparison of petroglyph presence on scoria bodies by sector. 85 
  
xi  
Table 20.  Comparison of closed curved form petroglyph presence on scoria bodies by 
sector. 687 
Table 21.  Estimated absolute scale values for dimensions of pukao, moai, and ahu at 
four locations used in sample physical calculations graphs. 114 
Table 22.  Estimated absolute scale values for ahu dimensions. 122 
Table 23.  Uncalibrated measurements between pukao features and centers of mass. 129 
Appendix Table 1.  List of past visiting groups with references to pukao. 136 
 
 
  
  
Introduction 
Rapa Nui (Easter Island) is a relatively young volcanic island that formed 
through hot spot volcanism starting around 2.5 million years ago (mya).  Between 2.5 
and 0.18 mya, a series of eruptions created Poike and Rano Kau, on the northeastern 
and southwestern corners of the island, respectively (Figure 1).  Starting approximately 
360 thousand years ago (kya), numerous lava flows from two main fracture systems 
created Terevaka, which currently dominates the geology of the island.  Lava flows that 
created Terevaka are quite young, some even dating to as late as 10 kya (Herrera and 
Custodio 2008, 1329).  The geological history of Rapa Nui has left it well endowed 
with stone resources such as basalt, andesite, scoria, and obsidian. 
 
Figure 1  Map showing terrain, major archaeological features, and infrastructure of 
Rapa Nui.  Source: “Easter Island.” Wikipedia: The Free Encyclopedia. Wikipedia 
Foundation, Inc. 10 Dec. 2011. Web. 20 May 2015. 
With an area of only 164 square kilometers and located over 2,000 km distant 
from the nearest inhabited land, Rapa Nui is truly both small and remote (Figure 2).  
 
 
2  
Polynesians first colonized the island in approximately 1200 AD and subsequently 
constructed monumental statues (moai) primarily from tuff, which they placed upon 
platforms (ahu) prior to European contact (Hunt and Lipo 2006).  Islanders additionally 
placed cylindrical bodies of red scoria (pukao) atop many of the completed moai.  The 
Rapanui1 quarried bodies of stone for moai and pukao at Rano Rarako and Puna Pau 
(Figure 3), respectively and subsequently transported them to ahu located primarily 
along the coast (Figure 4).  Based on obsidian hydration dating, Sue Hamilton (2013, 
100) suggests activity at the Puna Pau quarry between the 14th and 17th centuries (2013, 
100).  Unlike the tuffaceous moai, the pukao are made of a relatively soft and less dense 
scoria that geologist Mark Bandy (1937, 1605) describes as “coarsely vesicular, 
vesicles ranging up to over 15 cm in diameter.  The color varies from bright red, on 
weathered surfaces, to almost black on fresh fractures.  Much of it is ropy and 
scoriaceous.”  The density of the red scoria at Puna Pau is approximately 1.55 g/cm2 
(Edwards et al. 1996). 
                                                        1 Rapanui refers to the people of Rapa Nui. 
 
 
3  
 
Figure 2.  Map showing distances between Rapa Nui and nearest landmasses.  Satellite 
imagery provided by Google, SIO, NOAA, U.S. Navy, NGA, GIBCO, Landsat, and 
INEGI, 2015. 
 
Figure 3.  Photograph of red scoria quarry at Puna Pau (photo by author, 2014). 
 
 
 
4  
 
Figure 4.  Photograph of red scoria pukao atop moai on an ahu at Anakena, located on 
the northern side of the island.  Note that red scoria here is also used in the ahu lentils 
in front of the moai (photo by Terry Hunt, 2015).  
The Conventional Narrative 
The monumental statues on Rapa Nui, coupled with the lack of surface flowing 
water, sparse vegetation, and small population on the island, perplexed several early 
European visitors to the island.  To many, it seemed impossible for the natives’ former 
statue cult to thrive on an island so deprived of running water and forests.  Starting with 
the English expedition to Rapa Nui under Captain James Cook in 1774, European 
visitors began to theorize that the island was fully forested and had a more abundant 
water supply prior to European contact.  Specifically, naturalist Georg Forster believed 
that the Rapanui were “formerly more numerous, more opulent, and happy” (Kahn 
1968, 340).  To explain the current state of the island, French Captain Jean François 
Galaup de la Pérouse concluded in 1786 that the “imprudent” ancestors had destroyed 
 
 
5  
their freshwater and forest resources through total deforestation of the island (Passos 
1995, 55).  Many subsequent visitors to Rapa Nui adopted similar conclusions.  Most 
recently, Jared Diamond points to Rapa Nui prehistory as a case where islanders 
committed ecological suicide (termed “ecocide”) prior to European contact (2005).  
Diamond’s popular judgment of the island’s precontact past advances the historical 
trend of denying the past accomplishment of the Rapanui.  However, research by 
several archaeologists, led by Terry Hunt and Carl Lipo, has entirely undermined 
notions of ecocide on Rapa Nui (see Hunt and Lipo 2011).  Specifically, Diamond’s 
ecocide narrative fails to fully consider the effects of post-contact disease and slavery 
on the native population and overlooks the Rapanui’s sophisticated subsistence strategy. 
A key component in Diamond’s ecocide narrative involves moai transport.  
There exist several ideas that explain moai transport, but Diamond holds that the need 
for timber to fuel a statue building mania led to the total deforestation of Rapa Nui 
(2005, 118).  In a recent study of the form of moai, Lipo et al. (2013) establish that moai 
transport took place by walking erect statues and that this process did not require timber 
(Figure 5).  This discovery held significance both in dispelling the ecocide myth and in 
recognizing the ingenuity of the prehistoric Rapanui. 
 
 
6  
 
Figure 5.  Walking of a replica moai at Kualoa Ranch, O’ahu (photo by Terry Hunt, 
2012). 
Limited Historic Record of Pukao 
Despite the great interest in moai, few studies have investigated pukao 
production and transport.  Indeed, Bahn (1993, 84) notes that “The question of how 
these massive squat blocks were lifted meters off the ground and fixed accurately on the 
statue heads remains one of the most tantalizing enigma’s of Easter Island’s ancient 
culture.”  Rapa Nui has a turbulent history that has resulted in a loss of indigenous 
knowledge, so we must turn to the archaeological record in order to answer most 
questions about prehistoric monumentality.  The historical turbulence started with the 
likely introduction of disease with early European visits to the island and reached its 
peak during Peruvian slave raids in the mid to late nineteenth century (Fischer 2005, 
61).  The first slave raid came in 1805, when New England whalers aboard the ship The 
Nancy kidnapped 22 islanders (Charola 1994, 25).  Subsequent whalers visiting the 
 
 
7  
island likely introduced syphilis (Charola 1994, 25).  The largest slave raid to the island 
came in 1862, when eight Peruvian slaving ships anchored off Rapa Nui.  According to 
Thor Heyerdahl and Edwin Ferdon (1961, 67) fof the Norwegian Archaeological 
Expedition of 1955-56: 
“[The slave raiders] all decided to cooperate, and eighty armed men were 
sent ashore spreading trade goods on the ground to attract the natives.  
When about five hundred of the Easter Islanders were gathered, mostly 
on their knees examining the trade goods, the slave raiders fell upon 
them and captured two hundred, while nearly a dozen were shot dead…. 
Among those kidnapped were the island king, Kaimakoi, and his son 
Maurata, as well as nearly all the maori, or learned men, all of whom 
died on the guano islands.” 
Following international outcry against the brutal Peruvian slave raids, Peruvian slavers 
repatriated fifteen Rapa Nui.  However, those who returned “brought a smallpox 
epidemic to the rest of the island populations” (Heyerdahl and Ferdon 1961, 68).  
Overall, a series of population declines and the loss of the Rapanui intelligentsia led to a 
significant loss of indigenous knowledge by the time of the first ethnographic work on 
the island.  Though Wilhelm Geiseler (1882) and Katherine Routledge (1914-15) 
provide some relatively early ethnographic work, the majority of such work came a 
generation or more after the devastating slave raids and thus provide only partial insight 
into Rapa Nui’s past. 
Speculation on Pukao Meaning 
Given the loss of traditional knowledge surrounding pukao, the very meaning of 
the red scoria bodies atop moai is a source of debate.  Historically, visitors to Rapa Nui 
have variously referred to the red bodies of scoria that rest atop the heads of moai as 
crowns (Katherine Routledge), feather diadems (Alfred Métraux), turbans (Métraux), 
 
 
8  
headdresses (Linton Palmer), wigs (Everard Im Thurn), hats (Wilhelm Geiseler), 
millstones (Hippolyte Roussel), topknots (or pukao, Heyerdahl), and even baskets 
(Jacob Roggeveen).  Serious speculation as to the meaning of these scoria bodies began 
in the early twentieth-century with the visit of Katherine and Scoresby Routledge in 
1914-15.  After their visit, the two main ideas that emerged were that the scoria bodies 
represented either hats or hair, but the latter gained common use in the mid-twentieth 
century. 
Scoresby Routledge (1917, 334) first advances the hat interpretation by noting a 
parallel with “a hat very commonly worn made of native grass” that is “not dissimilar to 
that of the images.”  Advancing the hat interpretation, anthropologist W.H.R. Rivers 
asserts that the scoria bodies represent hats associated with “cult of the dead” (Rivers 
1920, 301).  He links the opposing hair interpretation with the idea that the prehistoric 
Rapanui fashioned them out of a desire for aesthetic beauty.  According to Rivers 
(1920, 301), religion, and not the drive for aesthetic beauty, is the only “motive strong 
enough to have led the makers of the statues to add these crowns and transmit the 
necessity for their manufacture to their descendants.”  He supports this speculation by 
noting the association of scoria bodies with burials on Rapa Nui and by drawing a 
parallel: “In the Marquesas a great stone is placed as a sign of mourning on the head of 
the image representing a dead man” (Rivers 1920, 301).  Rivers (1920, 304) also 
acknowledges the presence of headdresses elsewhere in Polynesia (e.g. in the Hawaiian 
Islands, Figure 6) that are not associated with rituals related to death.  H.D. Skinner 
(1922), of the University of Otago Museum, seeks to support Rivers’ interpretation of 
the Rapa Nui scoria bodies with a comparison to hats observed in New Caledonia 
 
 
9  
during Cook’s Second Voyage (Figure 7).  The writings of Rivers and Skinner appear to 
be the only ones following those of Scoresby Routledge that defend the hat 
interpretation. 
 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 6.  Engraving based on a 1778 sketch by William Ellis from Hawai’i during 
James Cook’s Third Voyage.  This depicts a ceremonial platform with figures that may 
be interpreted as wearing hats.  Source: “Heiau.” Wikipedia: The Free Encyclopedia. 
Wikipedia Foundation, Inc. 19 Feb. 2012. Web. 20 May 2015. 
 
 
 
 
 
 
 
 
 
10  
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Figure 7.  A 1777 engraving based on a sketch by William Hodges from New 
Caledonia during James Cook’s Second Voyage.  This depicts a man in a hat that H.D. 
Skinner compares to the pukao of Rapa Nui.  Source: “Man of New Caledonia.” David 
Rumsey Map Collection. 2005. Web. 20 May 2015. 
 
In contrast, archaeologist Henry Balfour (1921, 71) argues that the scoria bodies 
may be crude imitations of hair that is “lime-bleached after the Melanesian fashion.”  
Balfour (1921, 71) argues against the hat ideas by noting that Rivers’ connection with 
the Marquesas is too speculative to be taken seriously.  Métraux (1940/1971, 301) 
generally supports Balfour’s hair interpretation and specifically breaks down Rivers’ 
connection to the Marquesas by referring to the foremost researchers in the Marquesas, 
who found no evidence supporting the past use of stone crowns as a sign of mourning.  
 
 
11  
He notes that “we are by no means sure that the Easter Islanders ever bleached their hair 
with lime, and they certainly do not have the kinky hair of the Melanesians” (Métraux 
1940/1971, 301).  Instead, Métraux (1940/71, 301) bases his interpretation on hairstyles 
popular among the Rapanui: “The cylinder with the knob may have been an attempt to 
represent the long hair tied up on the head in a big knot (pukao), a fashion very common 
on Easter Island.”  Note that the possible mortuary significance highlighted by Rivers 
and the interpretation by Métraux are not necessarily incompatible, though it is difficult 
(if not impossible) to know the individual motivations of the builders of the pukao.  
Heyerdahl and Ferdon advance Métraux’s interpretation of the scoria bodies as pukao, 
and the idea that the scoria bodies represent the “islanders’ hairdo” is now firmly 
entrenched (Charola 1994, 23). 
Hamilton’s Rapa Nui Landscapes of Construction Project (2007, 2011, & 2013) 
describes the most recent work with pukao completed as part of a broader project 
examining the significance of Rapanui stone use in general.  Work by Hamilton et al. 
(2011) discusses the varied uses of red scoria and speculates on the past symbolic 
importance of stone to the prehistoric Rapanui.  Hamilton et al. (2011) and Van Tilburg 
(1986) both note the traditional association of the color red with the ceremonial and the 
sacred as a possible reason for the quarrying of vibrant red scoria from Puna Pau.  To 
highlight the significance of red scoria beyond use in pukao, Hamilton et al. (2011) note 
the prehistoric (and possibly historic) use of red scoria in ahu lentils (see Figure 4) and 
crematoria.  Regarding the link between red scoria and funerary rituals, Hamilton et al. 
(2011, 183) write, “crematoria associated with ahu contain spreads of small fragments 
of red scoria, mixed with pieces of white burned bone, dark flow lava and sometimes 
 
 
12  
obsidian flakes.”  Noting this association and the absence of red scoria from domestic 
structures, Thomas (2014, 104) also suggests “an association [of red scoria] with death 
and funerary rites,” which echoes Rivers’ interpretation of past pukao meaning.  
Overall, to Hamilton et al. (2011, 186), red scoria pukao production is part of a larger 
series of “material chains of signification from an incredibly monumental scale down to 
that of a grain of scoria in a cremation.” 
Building on Hamilton’s work, Thomas (2014) attempts to understand the 
distribution of pukao and why the Rapanui selected red scoria from Puna Pau for pukao 
production.  Thomas also provides an exhaustive list of the various uses and reuses of 
red scoria on Rapa Nui (Table 1) and a list of descriptions of the various red scoria rock 
outcroppings on the island.  Thomas (2014, 107) also speculates on the prehistoric 
governance of red scoria quarrying on Rapa Nui by noting a parallel:  “Analogous 
specialized use of stone from particular sources is closely paralleled on Hivaoa, in the 
Marquesas, where a quarry used for tiki and building stone used in sacred structures was 
governed by a strict tapu, and it is suggested that the use of Easter Island red scoria was 
controlled in a similar way.”  None of the research conducted by Hamilton or Thomas 
examines pukao production and transport, even though they address the distribution of 
red scoria and possible significance of various stone uses on Rapa Nui. 
 
 
13  
 
Table 1.  List of various uses and reuses of red scoria on Rapa Nui (after Thomas, 2014: 
96). 
Prior Thoughts on Pukao Transport  
Many early visitors to Rapa Nui marveled at the fact that multi-ton bodies of red 
scoria rested atop many of the moai.  However, most visitors have dedicated time to 
writing about, rather than investigating, the method of prehistoric pukao production and 
transport.  Historical sources agree that the Rapanui moved red scoria from Puna Pau to 
various points along the coastline by rolling it in relatively large cylindrical bodies.  The 
support for this comes from the fact that some of these relatively large cylinders are 
found at inland locations between Puna Pau and coastal ahu.  In particular, Katherine 
Routledge (1919/1998, 199) notes that “An unwrought cylinder is still lying at a 
hundred yards from the ahu of Anakena.”  William Mulloy (1970, 17) also recognizes 
this scoria body (Figure 8) and claims that it is about 150 meters inland of the nearest 
ahu.  
 
 
14  
 
Figure 8.  Image from Anakena facing W and giving a view of a large scoria body 
located at a distance from the nearest ahu (Ahu Ature Huke, photo by Terry Hunt, 
2015). 
 
Heyerdahl and Ferdon (1961, 373) also refer to two isolated specimens “far 
away from any ahu or statue, apparently abandoned during transport in the general 
direction of Anakena.”  While Lipo and Hunt (2005) have provided thorough 
documentation of moai transport roads, no one has yet attempted a mapping of pukao 
transport roads, likely due to lesser imprint left by the latter stone bodies.  Given that 
only a few red scoria bodies seemingly abandoned in transit remain, it is difficult to 
determine how the Rapanui rolled them in the past.  However, Mulloy (1970, 15) notes 
that “an encircling groove to be seen on several other topknots presumably in 
transportation suggests that the operation might have been assisted, at least in some 
cases, by the mechanical advantage of a cable rigged as a parbuckle.”  Overall, he posits 
that the Rapanui used a combination of pulling and levering as depicted in Figure 9. 
 
 
15  
 
Figure 9.  William Mulloy’s hypothesized method of prehistoric red scoria body 
movement on Rapa Nui (a-1) and notice of transition from circular to oblong section (a-
2, after Mulloy, 1970: 16). 
Opinions diverge when we come to how the massive scoria bodies are 
transformed into pukao and erected on moai upon reaching the ahu.  Generally, 
Heyerdahl and Ferdon (1961, 373) and Katherine Routledge (1919/1998, 199) share the 
idea that the prehistoric Rapanui created more refined pukao out of the larger scoria 
bodies only once the latter reached the ahu so as to reduce potential damage to the 
pukao.  Unlike the larger scoria bodies abandoned in transit, pukao include more 
delicate features such as cylindrical protrusions from the top and base indentations that 
fit the heads of moai.  Mulloy further supports the idea of pukao carving at ahu by 
contrasting the elliptical sections of the pukao with the relatively circular sections of 
larger scoria bodies (Figure 9).  Also note that Steadman et al. (1994, 85) found an 
exotic red scoria deposit (with an associated radiocarbon date of ca. A.D. 1220-1420) 
that they hypothesize to be “debitage from the final preparation of topknots” in a test pit 
just off the seaward side of Ahu Naunau at Anakena.  It is very likely that some type of 
recurving took place near ahu in order to create pukao. 
 
 
16  
 
Transport Mechanism Supporters Basis 
Rolling Up Ramp James Cook (1774) 
Hippolyte Roussel (1866) 
William Thomson (1886) 
Katherine Routledge 
(1914) 
Alfred Métraux (1940) 
Speculation 
Tradition 
Building Tower Beneath Thor Heyerdahl (1955) 
Vincent Lee (2012) 
Speculation 
Raising Pukao with Moai Thor Heyerdahl (1955) 
William Mulloy (1970) 
Speculation + Base Inden. 
Horizontal Moai Transport 
Wood Sliding Ramp Pavel Pavel (1995) Speculation and 
Experiment 
Zig-Zag Ramp Pavel Pavel (1995) Speculation 
Table 2.  Five proposed pukao transport mechanisms with their supporters and basis 
through time. 
What follows is a summary of the history of thought by visitors to Rapa Nui 
regarding pukao transport to the top of moai (see Appendix Table 1 for list of historic 
sources that include mentions of pukao).  As seen in Table 2, all of the ideas put forth 
include a degree of speculation.  Cook (1777, 295) provides the earliest speculation of 
pukao transport:  “a sort of mount or scaffolding would be made, upon which they 
might roll the cylinder, and place it upon the head of the statue; and then the stones 
might be removed from it.”  Cook acknowledges that this speculation is not grounded in 
any observations that he or his crew made during their stay on the island, yet the Ramp 
Hypothesis (Figure 10) reappears in the writings of many subsequent visitors.  It first 
reappears in vague terms in the writings of the French missionary Hippolyte Roussel, 
which refer to a traditional view that the pukao were “put on top of each statue by piling 
up stones” (Altman and Morin 2004, 498).  Writing twenty years later, William 
Thomson (1891, 499) of the Smithsonian expedition asserts without reference to 
 
 
17  
evidence that “The crowns were placed in position upon the heads of the standing 
images by building a road-way upon which they could be rolled to the proper spot.  The 
clearing away of the incline was the final act.  The earth which formed the surface was 
utilized as garden patches, and the stones which formed the foundation of the roadway 
were disposed of in building the wing-extensions of the platform.”  Katherine 
Routledge (1919/1998, 197) supports the Ramp Hypothesis by referring to a traditional 
story regarding a specific pukao on the north coast: “Near Paro, the ahu where the last 
statue was overthrown, there is a hillock, and tradition says that a causeway was made 
from it to the head of the tall figure which stood upon the ahu, and along this the hat 
was rolled.”  Métraux allegedly later encountered the more generalized ramp idea when 
an islander told him “that his ancestors had moved the pukao up onto the head of the 
moai by means of a pile of stones placed against the body of the figure” (Heyerdahl 
1989, 162). 
 
 
18  
 
Figure 10.  Cartoon depiction of a possible rendition of the Ramp Hypothesis (sketch 
by Roberto Foester for Pavel Pavel’s manuscript, obtained 2014). 
While many visitors subscribe to the Ramp Hypothesis and refer to traditional 
stories for support, this hypothesis came under criticism during the mid-twentieth 
century.  Specifically, regarding Routledge’s story of the pukao causeway, Heyerdahl 
and Ferdon (1961, 373) write that it “has the definite character of folklore” given “the 
impractically long distance to any nearby hill.”  Mulloy (1970, 15) elaborates on this 
impracticality: “To have begun at any nearby hillock such a causeway would have to 
have been at least 13 metres high and several hundred metres long - seemingly a 
prohibitive amount of construction.”  Heyerdahl and Ferdon (1961, 110) more generally 
note the impracticality of the ramp hypothesis by judging that moving a pukao up a 
 
 
19  
ramp is a “precarious undertaking for the number of men who could have gotten behind 
it.” 
In order to offer an alternative to the Ramp Hypothesis, Heyerdahl and Ferdon 
put forth the hypothesis that a pile of stones gradually stacked under a pukao may have 
raised it to the top of a moai (the Tower Hypothesis).  In this situation, the pukao rises 
“vertically with the bottom down by gradually building under it a cone of stones” 
(Heyerdahl and Ferdon 1961, 110).  The pukao is then levered laterally into place after 
it reaches the level of the moai.  As noted by Mulloy (1970, 16), “This would avoid the 
delicate problem of tipping the topknot” once it reaches the top of the moai.  However, 
based on assumptions of the top diameter of the conical pile (7-9 meters) and the angle 
of repose of the conical pile (6o degrees from the horizontal), Mulloy (1970, 16) 
concludes that this method involved a platform of “tremendous size” that would make 
this method “difficult, but possible.”  Vincent Lee (2012) later tries to improve the 
feasibility of this hypothesis by noting that the base spread of the pile of stones beneath 
the rising pukao could be reduced by reinforcing it with beams of timber (Figure 11).   
 
 
20  
 
Figure 11.  Depiction of the Tower Hypothesis, which involves a pile of stones built 
beneath the pukao (after Lee, 2012: 13). 
Heyerdahl and Mulloy offer a third hypothesis, which assumes that the moai and 
pukao are erected simultaneously as a unit (the Simultaneous Set Up Hypothesis, Figure 
12).  In this situation, wooden beams extending along the head of the moai and the 
length of the pukao are lashed together in order to affix the pukao to the top of the moai 
while both lie on the ground (Mulloy 1970, 16).  They are “then raised together by 
continually maintaining their relationship through adjustments in the single developing 
wedge of masonry under them” (Heyerdahl and Ferdon 1961, 110).  In 1998, Van 
 
 
21  
Tilburg demonstrated the physical feasibility of this transport idea by using it to erect a 
680 kg replica pukao with a 9 metric ton replica moai (T Ralston personal 
communication to C Lipo, 2015).  While this hypothesis effectively meshes pukao 
transport with the erection of a horizontally-transported moai, we now know that the 
moai were ‘walked’ erect to ahu (Lipo et al. 2013), which brings this hypothetical 
pukao transport method into question. 
 
Figure 12.  Depiction of the Simultaneous Set Up Hypothesis in which the moai and 
pukao are erected simultaneously (after Mulloy, 1970: 16). 
Believing in a vertical form of moai transport, engineer Pavel Pavel advances a 
fourth hypothesis for moai transport:  The Wooden Ramp Hypothesis.  He envisions 
two sloping beams of wood providing a ramp to the top of the moai, as shown in Figure 
13.  In this situation, levering and pulling brings the pukao to the top of the moai.  Pavel 
demonstrated the feasibility of this hypothesis by using it in with the assistance of four 
people in order to raise a 900 kg replica pukao along two spruce logs to the top of a 
replica moai at three meters height (Pavel 1995, 71).  Pavel’s discussion of pukao 
transport involves another speculative method:  The pukao is rolled side to side up the 
zig-zag path of stones in what is essentially a compressed version of the ramp 
 
 
22  
hypothesis.  However, Pavel does not provide evidence in support of this hypothesis, 
and it appears only in his writings. 
 
Figure 13.  Depiction of Wooden Ramp Hypothesis in which pukao is slid up a wooden 
ramp to the top of a moai (after Pavel, 1995: 71). 
Given Pavel’s successful experimental work with pukao transport in 1994, one 
might be tempted to say that he “solved the problem of raising a topknot on a statue 
long ago” (Love 2000, 116).  However, showing how the pukao could have been 
transported is not equivalent to showing how they were transported.  Neither Pavel nor 
Mulloy nor any previous visitors systematically examine the form of the pukao for 
evidence that may be uniquely indicative of one form of transport. 
Broad Significance of Research 
Thus far, I have briefly introduced the physical setting of Rapa Nui, the 
monumental statue building of the Rapanui, the loss of traditional knowledge on Rapa 
 
 
23  
Nui through slavery and disease, and the various ideas by visitors to the island 
regarding the meaning of pukao and possible methods of transport.  My research builds 
on existing knowledge by seeking to understand how the Rapanui produced pukao and 
placed them atop moai.  My project holds significance in drawing attention away from 
the association of pukao with ecocide, potentially changing our understanding of the 
role of pukao in prehistoric Rapanui survival, and preserving a record of existing pukao 
variation. 
Understanding pukao production and transport will lead to a better appreciation 
of the past technological accomplishment of the Rapanui: an appreciation that has been 
denied until relatively recently.  As the Orliacs (1995, 1) note, some people have even 
“refused to believe that the statues are the result of mere human capabilities.”  
Specifically, writer Erich von Däniken (1970) denies the achievements of the 
prehistoric Rapanui in monumental statue production by invoking highly speculative 
tales of extraterrestrials (Figure 14).  Meanwhile, Jared Diamond considers moai and 
pukao only within his popular ecocide narrative which judges the prehistoric Rapanui as 
imprudent.  Making a comparison between the pukao and modern mansions in order to 
support his judgment of the pukao as a sign of imprudence, Diamond (2005, 98) writes: 
“I cannot resist the thought that they were produced as a show of one-
upmanship.  They seem to proclaim: ‘All right, so you can erect a statue 
30 feet high, but look at me: I can put this 12-ton pukao on top of my  
statue; you try to top that, you wimp!’ The pukao that I saw reminded me 
of the activities of Hollywood moguls living near my home in Los 
Angeles, similarly displaying their wealth and power by building ever 
larger, more elaborate, more ostentations houses.” 
This speculation closely matches of that of Georg Forster, who believed that, prior to 
their modern downtrodden state, the Rapanui “could spare sufficient time to flatter the 
 
 
24  
vanity of their princes, by perpetuating their name by lasting monuments” (Kahn 1968, 
340).  Overall, in Diamond’s narrative (2005, 118), pukao and moai stand as remnants 
of a grim cautionary tale of human recklessness in what is “the clearest example of a 
society that destroyed itself by overexploiting its own resources.”  However, the moai 
no longer represent symbols of societal guilt in part since Lipo et al. (2013) 
demonstrated the ingenious method of their prehistoric transport, which did not require 
timber.  Comparable work with pukao may have a similar effect.  
 
Figure 14.  Writers such as Erich von Däniken, whether they intend to or not, often 
remove the agency of the prehistoric Rapanui in monumental statue production by 
invoking highly speculative tales of extraterrestrials.  Source: “Aliens Transporting 
Easter Island Moai Statues.” Easterislandtraveling.com. Web. 20 May 2015. 
Aside from modifying popular judgments of Rapa Nui monumentality, Hunt and 
Lipo’s research reveals how monumentality on Rapa Nui aided the survival of the 
Rapanui in a marginal and unpredictable environment.  For clarification, we must now 
 
 
25  
consider the ultimate value of monumentality in an evolutionary sense.  As O’Brien and 
Leonard (2001, 14) note, “It would make little sense to call a circle an adaptation, but it 
might make sense to call vessel decoration an adaptation within a given setting.”  
Similarly, it would make little sense to call the entity of a moai or pukao an adaptation, 
for they are not very handy tools in and of themselves.  However, producing and 
transporting moai and pukao may be explained in terms of adaptation.  Such an 
adaptation need not be aligned with the overt motivation for the prehistoric Rapanui in 
monument construction (e.g., possible desires to embellish moai with pukao for 
aesthetic, mortuary, and/or religious reasons).  Hunt and Lipo (2013, 17) recognize the 
evolutionary function of moai construction when they determined the method of moai 
transport: 
“The evidence for moai carving and transport points to activities by 
small-scale social groups rather than the product of laborers unified 
under a powerful central chiefdom.  Here monumentality does not imply 
large-scale social organization as assumed for many cases worldwide.  
Instead we see moai carving and walking as vivid expressions of costly 
signaling and evolutionary bet hedging in a competitive environment.  
Multiple lines of evidence, including the ingenious engineering to ‘walk’ 
statues, point to Easter Island as a remarkable history of success in a 
most unlikely place.” 
Bet hedging, a common strategy for stock investors (i.e., portfolio diversification), 
involves suppressing mean fitness in order to decrease variability in fitness through 
time.  This behavior holds evolutionary value in unpredictable environments such as 
that on Rapa Nui (e.g., Beaumont et al. 2009).  In the case of Rapa Nui monumentality, 
time spent with moai production and transport is a form of bet-hedging that takes time 
away from efforts leading toward rapid and unsustainable population growth (Hunt and 
Lipo 2011, 134).  Moai production and transport can also be considered an expression 
 
 
26  
of costly signaling, which holds benefit for the sender and receiver of an honest signal 
and evolutionary benefit for the population as a whole (Hunt and Lipo 2011, 132).  By 
linking moai to costly signaling and evolutionary bet hedging through an understanding 
of moai production and transport, Hunt and Lipo completely overturn Diamond’s 
interpretation of moai:  Instead of being symbols of imprudence and societal failure, the 
moai are remains of an evolutionary success story.  Determining the method of pukao 
production and transport has yet to be fully explained. 
Finally, given the ongoing degradation of pukao, my research provides baseline 
documentation that preserves the existing form of the pukao.  This parallels work by 
Kersten et al. (2009), which is part of a movement to start preserving the existing 
variation of moai through laser scanning.  The various factors at work in destroying 
pukao (and moai) through time will be discussed below. 
 
 
27  
Approach 
Bahn (1993, 84) notes that “Only experiments such as those already carried out 
in carving, moving, and erecting the moai can hope to enlighten us” when it comes to 
understanding pukao production and transport.  However, experimentation such as that 
by Pavel can only reveal what is feasible and what is not.  In my project, I adopt the 
approach of Lipo et al. (2013) in their study of moai transport and seek to understand 
pukao transport based on evidence in the archaeological record (i.e., to explain 
variability in that record, and not to speculate about plausible reconstructions alone). 
In this study, I treat the various aforementioned ideas on pukao transport as 
hypotheses to be tested through a systematic investigation of pukao variability.  Does 
the variability among pukao support one hypothesis above the others?  Does the extent 
of variability among pukao work against some hypotheses more than others?  In 
answering these questions, I build up from basic observations in order to test the 
hypotheses and the traits that we would expect to characterize each (Table 3). 
 
 
28  
 
  PUKAO WEAR 
 
 
 
 
 
 
 
GENERAL METHOD 
 
 
 
 
 
 
 
HYPOTHESIS 
D
ef
or
m
at
io
n 
on
 B
as
e 
H
or
iz
on
ta
l S
tri
at
io
ns
 
V
er
tic
al
 S
tri
at
io
ns
 
D
ia
go
na
l S
tri
at
io
ns
 
Si
de
 D
ep
re
ss
io
ns
 
 
Rolling and Tipping Ramp  X X  X 
Zig-Zag Ramp  X X X X 
 
Sliding Tower X X    
Wooden Ramp X X    
 
Neither Sliding nor Rolling Simultaneous Moai 
and Pukao Set Up 
  X   
Table 3.  The pukao transport hypotheses listed in Table 2 and expectations for 
associated wear.  If the Rapanui refinished pukao after erecting them, then the expected 
wear (except for that on the pukao base) would be erased.  Deformation on the base of 
pukao (including striations and irregular indentations) is expected as a product of 
sliding a pukao on its base.  Note that none, one, or a couple of the above methods may 
have been used by the prehistoric Rapanui. 
Evolutionary Framework for Understanding Style and Function 
I take an evolutionary approach outlined by Robert Dunnell in his discussion on 
style and function (1978) in order to understand pukao variability across space and 
through time.  Such an approach considers the archaeological record as a residue of the 
past differential persistence of various artifact traits in a cultural system.  I define 
dimensions as points of variation (e.g., striations present on the sides of pukao) and 
modes as types of variation within each dimension (e.g., vertical, horizontal, or diagonal 
striations).  Note that modes within some dimensions are on a continuous scale (e.g., 
length measurements).  The persistence of various modes depends on how the makers of 
the artifact in question interact with each other, with the artifact, and with the 
 
 
29  
surrounding environment.  These varied interactions give rise to a distinction between 
stylistic and functional variability. 
Modes within dimensions that are not under significant selective pressure are 
free to vary to a degree and are termed stylistic.  In the case of a series of wrenches 
(Figure 15), the dimension of composing material may be stylistic.  Though a metal 
with relatively high strength is required to maintain the integrity of the wrench, 
wrenches may be composed of various iron, copper, and aluminum alloys that are free 
to vary within reasonable limits.  Though often noted as neutral and inexplicable in 
terms of selection, stylistic variation can still be explained by evolution, explained 
below (Lipo and Madsen 2001, 304).  Additionally, because stylistic dimensions are 
transmitted across space and through time through the interactions of individuals and 
are unconstrained by selection, their propagation patterns are “accountable in terms of 
stochastic processes” (Dunnell 1978, 192). 
Figure 15.  A sample of wrenches illustrating stylistic modes that vary from wrench to 
wrench (e.g., composing metal) and functional modes that are shared by all (e.g. 
hexagonal notch).  Source:“Wrench.” The Typology. Diana Zlatonovski. 2010-2013. 
Web. 20 May 2015. 
 
 
 
30  
When a dimension comes under significant selective pressure, modes without 
selective value fall out of use and the overall variability within the dimension decreases.  
The modes that persist due to their selective value are termed functional.  In the case of 
the wrenches, the shape of the notch is constrained to hexagonal in order for the wrench 
to function properly.  All wrenches share this functional mode.  However, it is crucial to 
note that function, while it can be related to use, is not equivalent to use.  For example, 
the camouflaged color on a jet fighter is a functional dimension of the plane that is by 
no means equivalent to the use of the plane.  Also note that the term “functional” is 
limited in drawing a distinction from dimensions that are stylistic and those which are 
not.  This is because stylistic variation also has an evolutionary function in that it acts as 
a “reservoir of variability, some of which may ultimately acquire adaptive value with 
changing conditions” (Dunnell 1978, 199).  Returning to the example of the painted 
aircraft for clarification, note that “gray paint may be stylistic in peacetime, while 
serving as camouflage during combat” (O’Brien and Leonard 2001, 18).  Overall, a 
consideration of style and function reveals that “Both natural selection and stochastic 
processes may have a role in explaining cultural change” (Dunnell 1978, 198). 
Applications to Pukao 
I seek to distinguish functional modes of the pukao from stylistic modes.  The 
realization that function is not equivalent to use becomes key in considering possible 
functional dimensions in pukao.  The overt use of pukao from the perspective of the 
prehistoric Rapanui (e.g., for aesthetic, mortuary, and/or religious purposes) is 
ultimately unknown.  However, the functional detail (exhibiting relatively limited 
variability) is that which is constrained by the method of transport.  In the absence of 
 
 
31  
evidence to the contrary, I start with the assumption that pukao that are currently closely 
associated with moai once rested atop the nearby moai.  Thus, all pukao should share, or 
should have once shared, functional dimensions.  Such sharing of functional dimensions 
will test the hypotheses in Table 3. 
Distinguishing functional modes from stylistic modes will also allow for 
analysis of the variability of pukao that I deem to be stylistic.  A robust chronology 
linked to pukao transport does not yet exist, and the relatively small sample size of 
pukao across the island complicates the creation of a pukao seriation.  Martinsson-
Wallin and Crockford (2001) provide a summary of radiocarbon dates for the ahu of 
Rapa Nui, but many dates are based on unidentified charcoal fragments, and none are 
linked to pukao production.  In their assessment of radiocarbon dates in East Polynesia, 
Hunt and Lipo (2006) and Wilmshurst et al. (2011) identify large error stemming from 
in-built age in many of the radiocarbon samples used in the previous summary of dates.  
Relative assessments of pukao age can be based on condition, but weathering of pukao 
is determined by a multitude of poorly understood variables aside from time (e.g. 
distance from sea spray, lichen and other biological growths and exposure to wind, 
sediments transported by wind, and water).  Relative to moai production, we might 
assume that pukao came late.  This stems from reasoning first voiced by Métraux 
(1940/1971, 302): “If both devices had grown up simultaneously, the use of the big hats 
would have been more general and would not have been restricted to certain ahus or to 
a few statues on an ahu.”  Heyerdahl and Ferdon argued that, while relatively old moai 
“have rounded heads and probably were not designed to bear topknots,” younger ones 
include flat heads that allow for easier pukao balance (1961, 215). 
 
 
32  
The absence of pukao dating and limited sample size make it difficult to trace 
the stochastic processes of style transmission through time in pukao.  However, I expect 
that stylistic modes can be distinguished from functional ones in the existing sample 
based on the limited distribution of stylistic modes versus functional modes (Figure 16).  
In path A of Figure 16, a mode gradually becomes widespread under selective pressure 
likely due to functional value that it has in a given selective environment.  When this 
environment changes and the function of this mode is lost, the mode ceases to be 
transmitted.  In path B, a mode does not come under selective pressure, does not 
become widespread, and is not termed functional.  In path C, a mode comes under 
selective pressure, becomes widespread, and is termed functional.  Note that this 
interpretation does not consider preservation bias and the possibility of the rise in a 
homogenous style without functional value. 
 
Figure 16.  Hypothetical changes in frequency of modes under selection versus modes 
under drift (after O’Brien and Leonard, 2001: 9). 
The present frequency of modes that we observe in the archaeological record (at 
the endpoint on the time axis in Figure 16) is simply an integral of each mode frequency 
 
 
33  
over time.  I thus consider time as an interval out of necessity and ignore the “noise” in 
mode frequencies within the given interval.  By comparing the behavior of stylistic and 
functional modes in Figure 16, we quickly find that the two can be distinguished based 
on their relative abundances.  Overall, given our assumption of all existing pukao at one 
point being located atop moai, all pukao should share the functional modes.  Modes that 
are not shared by all pukao will be due to stylistic variability in the absence of other 
confounding variables such as differential preservation. 
 
 
34  
Methods 
Model Generations 
In order to locate and document the pukao of Rapa Nui during my visit to the 
island between May 20th and June 9th, 2014, I relied on a database of pukao surveyed 
with GPS in previous years by teams under Drs. Hunt and Lipo.  My documentation 
consisted of using a couple of Nikon cameras to take approximately 15,000 photographs 
with embedded GPS coordinates.  Using these photos and structure from motion 
mapping, I generated three dimensional computer models with embedded scale of 53 of 
the 75 pukao and 13 of the 23 Puna Pau scoria bodies identified in the existing database.  
In the pages that follow, I refer to specific scoria bodies by their assigned number in the 
existing database.  I also worked with Dr. Lipo in recording the physical context of 
pukao at a few ahu (at Vaihu, Ura Uranga te Mahina, and Akahanga, on Rapa Nui’s 
southern coast) by collecting aerial photographs with a DJI Phantom quadcopter and 
using these photos to develop orthophotos through structure from motion mapping. 
Structure from motion mapping is an inexpensive form of computerized 
photogrammetry through which three dimensional structure can be obtained from large 
sets of overlapping images.  The structure from motion software that I use (Agisoft 
Photoscan 1.0.4, 2014, Agisoft LLC, St. Petersburg, Russia) generates point clouds and 
meshes by picking out and matching hundreds of thousands to millions of unique points 
among photos (see section on “Model Generation, Filing, and Observation Strategies” 
in Appendix).  Figure 17 below gives a simplified diagram for visualizing the method 
behind structure for motion mapping. 
 
 
35  
 
Figure 17.  Illustration of the method behind structure for motion mapping.  Source: 
“Globally Optimal Structure and Motion Estimation.” UCSD Computer Vision. UCSD 
School of Engineering. 10 Nov. 2010. Web. 20 May 2015. 
When applying photogrammetric mapping to objects, it is important that images 
cover the entire surface and converge on the object from a close range.  The number of 
photos required to generate a model depends on the size, shape, and texture of the 
object, but a complete pukao model typically requires 50 photos at a very minimum.  
When created properly, these models preserve color and have sub-centimeter accuracy 
(Figure 18).  In addition to producing highly accurate and inexpensive models that can 
be easily measured and shared, structure from motion mapping is useful given that it is 
minimally invasive and fast. 
 
 
36  
      
Figure 18.  One of 290 photos (left) used to generate a textured model of pukao 18 at 
Vaihu in Photoscan (right). 
Recording Strategy 
Following the completion of model generations, I recorded surface variations 
among pukao in an open source model-viewing software called Meshlab (Figure 19).  A 
publication by Mudg et al. (2010) provides useful guidelines on how to most effectively 
identify surface variation of artifacts using three-dimensional imaging.  For my work, I 
used an APSS filter that colorizes curvature.  Adjustments in this filter can be made in 
order to modify the curvature measurement method and the scale of curvature 
measurement.  Applying these filters reveals detail previously obscured by shadow or 
surface color differences.  
 
 
37  
 
Figure 19.  Textured model of pukao 18 colored in Meshlab according to the APSS 
filter in order to highlight surface details. 
After a preliminary examination of pukao variability, I identified various 
dimensions of variability and the modes within each dimension.  Dimensions with 
categorical modes are given in Table 4, while those with modes that are continuous 
variables are listed below: 
• Volume 
• Maximum body and top cylinder diameters 
• Minimum body and top cylinder diameters 
• Body and top cylinder heights 
 
 
38  
 
Table 4.  Dimensions and modes of categorical variables. 
I discuss the details of each dimension and mode in subsequent sections.  
Currently, it is important to recognize that modes within each dimension are all 
inclusive and that this stipulates observations for use in paradigmatic classifications.  
Paradigmatic classifications of pukao will enable me to evaluate which surface features 
are stylistic with related temporal and spatial variability and which are functional and 
relate to production and transport.  
Paradigmatic classifications are an expedient tool to examine pukao variability.  
I share the view that “Classifications, or measurement scales, are however, devices of 
science that are imposed on, rather than extracted from, empirical reality.  They function 
Dimension Mode Dimension Mode 
Resting Face  Striations Present Along: 
B Base H Horizontal  
T Top V Vertical   
S Side D Diagonal only 
A Atop Moai HV Horiz. and Vert. 
BS Base and Side HD Horiz. and Diag. 
TS Top and Side VD Vert. and Diag. 
  HVD Horiz., Vert., and Diag. 
Condition  0 Not present 
P Poor   
F Fair Cupules Present On:  
G Good B Base 
E Excellent S Sides 
  T Top 
Exposure  BS Base and Sides 
1 Unburied BT Base and Top 
0 Partially Buried ST Sides and Top 
  BST Base, Top, and Sides 
Top Cylinder  0 Not Present 
1 Present   
0 Absent Other Surface Markings  
U Unknown Cu Isolated Curved Lines 
  Ci Circles 
Base Indentation  Cl Closed Curved Forms 
1 Present S Slab Cut Away 
0 Absent B Basin Dug 
U Unknown W Western Text 
  E Englert Number 
General Form  x,x,x Combinations of Above 
CY Cylindrical   
CO Conical   
 
 
39  
to provide the terms by means of which we identify, describe, measure, and compare 
phenomena; in short, they allow us to model reality in particular ways for particular 
purposes” (Dunnell 1986, 152).  Unlike definitions for empirical units, artifact 
classifications define theoretical units (Dunnell 1986, 152).  These latter units can be 
manipulated to fit our needs, and paradigmatic classifications provide a convenient 
means of gaining this flexibility.  Paradigmatic classifications are ones in which the 
intersection of various attributes form classes.  Melinda Allen (1996, 101) gives the 
following requirements for paradigms in such a classification: 
• each class is defined by the same set of criteria: if size is an attribute, then 
it is considered for all types not just some; 
• the attributes are not weighted: no feature is considered more important 
in type separation than any other; 
• the modes are mutually exclusive: only one value can be displayed at a 
time; and 
• the modes are exhaustive: one value must be displayed 
Allen notes that classifications founded on this basis are “a particularly parsimonious 
means of creating analytic units” (1996, 101).  Paradigmatic classifications can be 
easily modified in order to accommodate a potentially infinite degree of variability. 
In treating dimensions with continuous modes, I avoid creating arbitrary 
categorical units based on ranges of numbers and forcibly applying paradigmatic 
classifications.  I instead systematically examine the spread of these continuous modes 
and their spatial distribution. 
 
 
40  
Results 
Sample Description 
The quarry at Puna Pau includes scoria bodies, while finished pukao rest in 
clusters around ahu along the coastlines of Rapa Nui.  Most locations that yielded 
usable models are along the southern coast (Figure 20).  Only two locations on the 
island are not represented with models of the scoria bodies present: an unnamed 
location on the southern coast and relatively remote Vaimata on the northwest coast of 
the island (Figure 21).  Most scoria bodies that were not modeled are from Anakena and 
Puna Pau. 
 
Figure 20.  Map showing locations on Rapa Nui that yielded usable models of scoria 
bodies (green) and locations that include only pukao fragments that were not modelled 
(red).  The panchromatic image used in this map is one collected by DigitalGlobe’s 
QuickBird satellite in September, 2005.  This image is projected according to the World 
Geodetic System (WGS) 1984 datum for Universal Transverse Mercator Zone 12S. 
 
 
 
41  
Model Development Summary
Ta
ha
i
Vi
na
pu
 1
Vi
na
pu
 2
Ha
ng
a P
ou
ku
ra
Ta
ra
kiu
Va
ihu
Ur
a U
ra
ng
a t
e M
ah
ina
Ak
ah
an
ga
So
uth
er
n C
oa
st
Tu
'u 
Ta
hi
On
e M
ak
ihi
To
ng
ar
iki
He
kii
An
ak
en
a
Te
 P
ito
 K
ur
a
Ov
ah
e
Va
im
ata
Pu
na
 P
au
N
um
be
r o
f S
co
ria
 B
od
ie
s
0
5
10
15
20
25
Models Developed
No Models Developed
 
Figure 21.  Graph showing the number of models generated from each location (black) 
and the number of models at each location that were not modeled (grey).  Stacked bars 
give the total number of scoria bodies present at each location. 
Most were not modeled due the fact that they were buried, fragmentary, and/or 
difficult to locate (Table 5).  In most cases, heavily eroded fragments of red scoria that 
clearly composed less than half of an original pukao were deemed unusable.  There 
were only a few cases in which complete pukao were not modeled.  Specifically, pukao 
66 and 67 on the restored ahu at Anakena could not be effectively modeled because 
overlapping photographs could not be taken of their sides, which are obscured from a 
ground perspective by the adjacent restored pukao. 
 
 
 
 
 
42  
Pukao Number(s) Location Reason(s) 
11,12,13 Hanga Poukura Fragments buried, missing 
20,22,24 Vaihu Fragments missing 
30,31 Ura Uranga te Mahina Almost entirely buried 
40 No Data N/A 
42,43 Southern Coast Fragment buried, missing 
45,46,47 Near One Makihi Fragments missing 
63,64 Anakena Mostly buried, limited access 
66,67 Anakena Incomplete coverage due to 
restoration  
69,70,72 Anakena Extremely eroded fragments, 
some missing 
75 Vaimata Almost entirely buried 
Nine scoria bodies Puna Pau Limited access, tall grass, 
burial 
Table 5.  Summary of explanations as to why scoria bodies in given locations were not 
effectively modelled. 
Issues of Preservation 
Numerous processes degrade pukao through time and thus limit the current 
sample size.  These factors can be divided into those due to humans, the elements, and 
non-human organisms as described below.  Pukao are especially susceptible to 
weathering because they are composed of relatively soft and highly vesicular scoria. 
Human Factors 
Pukao currently on the ground once rested atop moai, so the first major source 
of damage came with the toppling of moai.  The role people may have played in 
toppling moai is unclear.  Roussel gives the earliest reports of moai toppling through 
warfare in the mid-nineteenth century, but few of Roussel’s contemporaries mention 
such warfare (Edwards et al. 1996, 13).  However, Scoresby Routledge (1917, 335) 
recounts an oral tradition in which “The last one [moai with pukao] was overthrown by 
the fathers of the present old men who as boys assisted their fathers – that is, about 
 
 
43  
1835.”  This relatively large moai (named Paro) and its relatively intact pukao lie near 
the ahu at Te Pito Kura on the northern coast of the island.  The fallen pukao, such as 
that belonging to Paro, have remained on the ground ever since, although five (four in 
Anakena and one in Tongariki) have been restored in the past fifty years.  Prior to 1835, 
visitors to Rapa Nui noted both toppled and erect pukao.  Even during Cook’s 1774 visit 
to the island, Johann Forster noted a fallen pukao: “We reached the east side of the 
island, near a range of seven pillars or statues, of which only four remained standing, 
and one of them had lost its cap” (Kahn 1968, 336). 
Once on the ground, islanders reused the red scoria from pukao to some degree 
(see Table 1), and this continued to erode the surviving pukao.  Note that it is difficult 
to trace the primary and secondary uses of stone from pukao given the varied uses of 
red scoria through time.  Regardless, as Thomas (2014, 106) sensationally puts it, “Over 
the years, red scoria originally from Puna Pau has been systematically robbed from ahu, 
with topknots cut and gouged away and granules from these spilled like blood over the 
ahu and around the heads of fallen moai.”  Katherine Routledge (1919/1998, 199) is the 
first to note reuses of pukao “as building materials.”  Thomas (Figure 21, 2014, 107) 
suggests the possible reuse of red scoria in inhumation burials and mentions a pukao 
from the vicinity of Tahai that was incorporated into a cross in the Catholic cemetery in 
Hanga Roa.  Native Rapanui also informed an archaeological team in the 1960s that 
another pukao in the vicinity “had been cut up to use as building material” (Charola 
1994, 35).  Many headstones in the cemetery are also made of a red scoria ultimately 
from Puna Pau, but it is unclear as to whether or not pukao scoria also contributed to 
these headstones.  The single pukao that currently rests atop a restored moai at Tahai is 
 
 
44  
a replica installed between 1968 and 1970 (Figure 22).  As Charola (1994, 35) recounts, 
“a replica pukao was set on top of the moai, with the intention of removing the topknot 
after the photographic coverage for an article about the island was completed.  But the 
local inhabitants objected, requesting that the pukao be left on the statue, where it 
remains.” 
    
Figure 22.  Red scoria reused in Catholic cemetery in Hanga Roa (left, after Thomas, 
2014: 106) and replica pukao at Tahai (right, photo by author, 2014).  Note that the 
replica is not considered in any subsequent analysis. 
Aside from toppling and subsequent reuse, many scoria bodies have western text 
engravings (graffiti).  However, the damage due to graffiti is minor in comparison to the 
damage due to red scoria reuse.  Most of the graffiti occurs in Puna Pau, and its age is 
unclear.  In 1935, Rapa Nui was designated as both a historic monument and a national 
park.  However, it was not until 1966 that the Chilean government assigned official 
personnel to the island (Charola 1994, 51).  Rapa Nui became a UNESCO World 
Heritage Site in 1995.  Official supervision may dissuade graffiti, but it is still possible 
that much of the graffiti may be quite recent. 
 
 
45  
The Elements 
During his visit in 1868, Linton Palmer (1869, 374) noted that pukao “were 
much decayed by weather.”  By the time of his visit, the pukao had already experienced 
heavy erosion due to a variety of physical and chemical processes independent of 
human action.  Rapa Nui’s coastlines are exposed to the winds, and the island 
experiences an average of 1,130 mm of rainfall each year, mostly in the form of 
showers (Charola 1994, 15).  It follows that “Heavy showers combined with strong 
winds erode the stone mechanically” (Charola 1994, 45).  In this process, the red scoria 
weathers to form clay minerals.  These weathering products settle in the interstices of 
the porous scoria and significantly expand when wetted and contract when dried (Figure 
23).  Accordingly, they create internal stress that aids the breakdown of the rock.  Note 
that the crystallization of salt from sea spray that blows into pukao can have a similar 
affect: one that is commonly referred to as salt wedging. 
 
 
46  
 
Figure 23.  Diagram illustrating expansion of a clay mineral with a sheet structure (after 
Charola, 1997: 29).  Note that water is drawn in between the weakly bonded sheets by 
attraction between the slight negative charges in the sheets of silicate tetrahedral and 
the slight positive charges in the hydrogens on the water molecules. 
 
In a few cases, intertidal wave action has significantly eroded pukao.  A scoria 
body that might be a  possible pukao near Ovahe (pukao 74) and several pukao from 
Vaihu have been subject to this form of erosion (Figure 24).  Spanish visitors to the 
island in 1770 noted the presence of several pukao in the intertidal zone, and this led 
them to conclude that the “stone of which these columns are made is not native to the 
island.” (de Haedo 1908, 74).  It was not until Cook’s visit to the island four years later 
that visitors recognized Puna Pau (“a small hollow…with several such cylinders as are 
placed on the heads of the statues”) as the source of red scoria used in pukao production 
(Ruiz-Tagle 2005, 163).  The fact that the Spanish noted pukao in the intertidal zone 
 
 
47  
indicates that these pukao likely underwent intertidal erosion for over 150 years.  The 
Norwegian Archaeological Expedition rescued the pukao at Vaihu from the intertidal 
zone in 1955.  However, these pukao nonetheless exhibit noticeably ablated surfaces. 
 
Figure 24.  Photograph of pukao in intertidal zone at Vaihu prior to salvage efforts 
during Norwegian Archaeology Expedition in 1955 (after Heyerdahl, 1989: 39).  Note 
that pukao 18 is at center, and that abrasion can be seen toward the base of this pukao in 
Figure 18. 
 
In addition to the slow, persistent onslaught of the elements, cataclysmic natural 
events such as earthquakes and tsunamis have also aided the breakdown of pukao.  
Rapa Nui is located in an active volcanic zone, and recordings from the seismologic 
station built on Rapa Nui by the Department of Geophysics of the University of Chile in 
1950 reveal that “earthquakes on Easter Island are not isolated events” (Edwards et al. 
1996, 2).  Edwards et al. examine the stability of ahu and conclude that much of the 
structural damage in ahu seen today can be explained by earthquakes.  We have direct 
evidence that at least one tsunami significantly damaged ahu.  Specifically, in 1960, 
Ahu Tongariki absorbed the brunt of a tsunami that strew moai, pukao, and ahu 
 
 
48  
fragments across an area of 1.4 hectares (Charola 1994, 37).  Figures 25 and 26 depict 
Tongariki before and after the tsunami.  Wind, rain, clay and salt wedging, earthquakes, 
and in some cases tsunamis and intertidal erosion have all worked to destroy pukao over 
the years. 
 
 
49  
 
 
 
Figure 25.  Photographic panorama of the front of the ahu showing fallen moai and 
pukao at Tongariki in 1914-15 (after Routledge, 1919/1998: Figure 34). 
 
 
 
Figure 26.  Photographs of restored ahu at Tongariki with a single moai with restored 
pukao (top) and nearby alignment of pukao following the 1960 tsunami (bottom, photos 
by author, 2014). 
 
 
 
50  
Non-Human Biological Factors 
Organisms, such as algae, lichens, sheep, cows, and horses have also played 
roles in the breakdown of pukao.  Charola (1994, 45) provides a concise description of 
the roles of algae and lichens: 
“These organisms can have a dual action on the stone: they selectively 
leach out irons, thus weakening the stone matrix, and/or induce 
mechanical stress by retaining moisture and contributing to the wet/dry 
cycles of the stone.  In the case of lichens, the penetration of rhizines or 
hyphae causes a mechanical stress that contributes significantly to the 
detachment of flakes from the stone surface.” 
Lichens are particularly good at obscuring and destroying petroglyphs through time 
given that they have the effect of smoothing edges (Figure 27).  Little conservation has 
focused on removal of lichen from pukao given that it is incredibly difficult to remove 
lichen without crumbling the stone into which its rhizomes have penetrated. 
 
Figure 27.  Pukao 55 at Tongariki covered in lichen (photo by author, 2014).  Note that 
lichen is also present on the pukao in Routledge’s panorama of Tongariki in Figure 25.  
This indicates that lichen has been a contributing factor in pukao degradation for over a 
hundred years.  Generally, lichen appears in greater abundance on pukao located in the 
northeastern portion of the island. 
Sheep played a role in the destruction of pukao when sheep ranching (primarily 
organized by the Compañía Explotadora de Isla de Pascua) dominated the island 
 
 
51  
following the annexation of Rapa Nui by Chile in 1888 and until the island was 
designated as state property in 1933 (Charola 1994, 27).  During this period, the ranch 
administrators confined the native Rapanui to Hanga Roa and gave sheep free reign of 
the entire island (Figure 28).  The sheep may have damaged intact pukao, but they 
likely had a greater effect in trampling and breaking down smaller fragments of broken 
pukao.  Today, grazing animals with access to some pukao continue to have damaging 
affects. 
 
Figure 28.  Undated photograph of sheep roaming Rapa Nui (photo by J. Douglas 
Porteous, 1981). 
Overall, these inorganic and organic processes hugely limit the number and 
condition of intact pukao that we see today and highlight the importance of maintaining 
a record of remaining pukao variation.  In my analysis of pukao variability, I assigned a 
qualitative condition mode to each pukao based on the completeness and symmetry of 
 
 
52  
each (Figure 29).  Note that pukao at Anakena include most of the pukao in excellent 
condition but that mode abundances are quite mixed at other locations. 
 
 
Figure 29.  Chart illustrating the number of scoria bodies within each condition mode 
and representative samples from the extreme modes.  Pukao 21 is in poor condition, is 
from Vaihu, and is seen in the upper left corner of Figure 24 prior to its removal from 
the intertidal zone.  Pukao 66 is in excellent condition and is from Anakena (photos at 
right by author, 2014). 
Note on Terminology 
In analysis of pukao variability, there exist two red scoria bodies included by 
others in the pukao database and not located at Puna Pau that I do not consider to be 
pukao (i.e., finished form).  Generally, my criteria for accepting a body of scoria as a 
pukao is that it is cylindrical or conical and closely associated with moai at an ahu.  I 
refer to red scoria bodies that do not fit this criteria and are not ahu lentils as merely red 
scoria cylinders or bodies of red scoria.  When pukao and other cylindrical red scoria 
 
 
53  
bodies are considered together, I refer to the assemblage as one of red scoria bodies.  
Two bodies of scoria are exceptionally large and located at quite a distance from moai 
(Figure 30). 
   
Figure 30.  Photographs of the two relatively large cylindrical bodies of red scoria 
included in the pukao database and not located in Puna Pau that are not considered to be 
pukao (photos by author, 2014).  Scoria body number 71 (left) is located about 150 
meters distant from the nearest ahu at Anakena.  This scoria body was in transport and 
can be considered unfinished, and thus not a pukao. Scoria body 62 (right) is located 
about 200 meters distant from the nearest ahu at Hekii. 
Volume and Weight 
Photoscan software enables measurements of volume from models with 
embedded scale, but a variety of factors limits the useable number of volume 
measurements in this study.  First, measurements of red scoria body volume are limited 
to those with full exposure.  Of the models generated, about 80% (42 of 53) are of fully 
exposed bodies.  In a few cases, limited access in the field inhibited accurate volume 
measurements.  For example, coarse models of a few restored pukao (56 at Tongariki 
and 66 and 67 at Anakena) inhibit useful volume measurements.  A problem also exists 
in the embedded scale of models of scoria bodies photographed from afar (e.g. pukao 44 
at One Makihi, which has an embedded scale that is multiple orders of magnitude off).  
Though the scales in such models do not correspond correctly to an absolute scale, they 
 
 
54  
are internally consistent.  Most models include some skew in embedded scale, and I 
calibrated these values by using field measurements taken of pukao under the lead of 
Drs. Hunt and Lipo in previous years.  Unfortunately, only 16 pukao volume 
measurements had associated average errors of less than one hundred percent. 
Plotting the remaining 16 pukao volume estimates against distance from quarry, 
we find a slight positive correlation (Figure 31).  This general increase in volume away 
from the quarry is also coupled with a relatively sharp increase in the number of red 
scoria cylinders after passing beyond an 11 kilometer radius of the quarry (Figure 32).  
It is only after passing beyond an 11 kilometer radius that pukao-bearing ahu located on 
the northern coast come into view and contribute to the count of those located on the 
southern coast. 
Pukao Volume as a Function of Distance from Quarry
Distance from Quarry (km)
2 4 6 8 10 12 14
Vo
lu
m
e 
(m
3 )
0
2
4
6
8
10
y = 0.2592x + 1.2268
r2 = 0.344
 
Figure 31.  Volume of pukao plotted as a function of distance from quarry.  Note the 
insignificant positive correlation. 
 
 
 
55  
Numerical Distribution of Red Scoria Cylinders Relative to Quarry
Distance from Quarry (km)
0 2 4 6 8 10 12 14
C
um
ul
at
e 
N
um
be
r o
f R
ed
 S
co
ria
 C
yl
in
de
rs
0
10
20
30
40
50
60
70
 
Figure 32.  Cumulative number of red scoria cylinders as a function of distance from 
quarry.  Circles of larger radii around the quarry include larger numbers of red scoria 
cylinders.  The horizontal line marks half of the cumulative total.  Note the relatively 
sharp increase in the number of red scoria cylinders after passing beyond an 11 
kilometer radius of the quarry. 
Given the density of red scoria, determined by Edwards et al. (1996) to be 1550 
kg/m3, we can calculate the masses of the well-calibrated pukao.  Unfortunately, only 
large ranges of possible mass can be gleaned from the calibrated data.  The least 
massive pukao in Figure 31 is from Hanga Poukura (pukao 14), and it has a mass 
anywhere between 900 and 2,400 kilograms (kg).  The largest pukao in Figure 31 is 
from Tongariki (pukao 49), and it has a mass anywhere between 7,600 and 14,000 kg. 
The most massive pukao (pukao 73, Figure 33) is at Te Pito Kura, belongs to Paro, and 
has a mass of about 11,500 kg (Bahn 1993, 84).  Overall, pukao appear to range in mass 
between one and almost twelve metric tons. 
 
 
56  
 
Figure 33.  Photograph collected by Métraux of the pukao belonging to Paro at Te Pito 
Kura (after Métraux, 1940/1970: Appendix). 
Base Indentations 
When the Spanish visited the island in 1770, they marveled at how pukao 
balanced atop moai.  F. A. de Agüera y Infanzon (1770/1908, 95) notes, “the diameter 
of the crown is much greater than that of the head on which it rests, and its lower edge 
projects greatly beyond the forehead of the figure; a position which excites wonder that 
it does not fall.”  González y Haedo satisfied their curiosity about how this balance is 
possible by noting the presence of base indentations in the pukao: “I was able to clear 
up this difficulty on making an examination of another smaller statue from whose head 
there projected a kind of tenon, constructed to fit into a sort of slot or mortice 
corresponding to it in the crown; so that by this device the latter is sustained 
notwithstanding its overlapping the forehead” (Ruiz-Tagle 2005, 58).  Few subsequent 
visitors made further note of pukao base indentations, but Heyerdahl and Ferdon (1961, 
372) write that, “this depression is not always in the center, but often slightly to one 
 
 
57  
side, so that the cylinder projected some distance out from above the statue’s 
eyebrows.” 
Of all of the scoria bodies observed in the field outside of Puna Pau, 42 are 
indeterminate when it comes to the presence or absence of base indentations.  This is 
accounted in part by poor preservation and exposure but is primarily because many rest 
on their base, partially buried on their sides, or atop moai so that we cannot immediately 
determine whether or not they include base indentations (Figure 34).  All ten of the 
pukao with bases visible include base indentations, and they are located throughout the 
island (Vinapu, Vaihu, Akahanga, One Makihi, and the Vicinity of Te Pito Kura).  In all 
cases, base indentation depths are only a few centimeters or less.  Pukao 15, which rests 
on its side at Hanga Poukura, does not appear to include a base indentation despite its 
good condition and reasonable exposure.  However, note that there is a slight concavity 
on the base of this pukao that may be the weathered remnants of a more clearly defined 
base indentation.  Overall, despite the limited number of observable base indentations 
and the exception of the pukao at Hanga Poukura, pukao across the island appear to 
share base indentations. 
 
 
58  
 
Figure 34.  Chart illustrating the number of scoria bodies within each resting face mode, 
and representative samples of a few of the modes.  At right, going from top to bottom, 
are pukao 15 from Hanga Poukura resting on its side, pukao 57 from Hekii resting on 
its side and top, and pukao 49 from Tongariki resting on its base (photos at right by 
author).  Note that most scoria bodies with indeterminate resting face are in poor or fair 
condition.  Most indeterminacy stems from confusion over whether a pukao is resting 
on its base or top. 
We can gain some understanding of how pukao balanced atop moai given base 
indentations as well as the centers of mass of these pukao.  Specifically, note that pukao 
are unstable when they rest atop moai and their center of mass is situated outside of the 
footprint of the base indentations (Figure 35).  I calculated the centers of mass of pukao 
models in Meshlab.  Two of the ten pukao with clearly visible base indentations, both of 
which are located at Hekii, have reasonable condition and full exposure that enable 
 
 
59  
useful center of mass measurements (Table 6).  Table 6 also includes center of mass 
measurements from two pukao resting atop moai at Anakena that may have base 
indentations but lack measurement calibrations. 
Overall, note the variability in how close the two pukao at Hekii are to tipping 
due to horizontal instability.  In pukao 57, the center of mass is 23 ± 2 cm from the edge 
(about 14% of total width of pukao) and in pukao 60, it is just 19 ± 3 cm from the edge 
(about 43% of total width).  The center of mass lies between 40 and 50 percent up the 
side of the pukao considered.  Given the position of the center of mass relative to the 
base indentations of the pukao, we find that the indentations are not necessary for 
balance once the pukao are atop moai. 
 
d = horizontal distance between center of mass (CM) and forward edge of base indentation 
w = minimum width of pukao 
h = vertical distance between base of pukao and CM  
p = total height of pukao 
 
Figure 35.  Sketches of base (left) and side (right) views of pukao and the variables 
considered.  Note that the pukao will be unstable while resting horizontally atop a moai 
if its center of mass lies outside of the base indentation (shaded) footprint in the left 
sketch. 
 
 
60  
 
Location Pukao Number d (cm) h (cm) d/w h/p 
Hekii 57 23 ± 2 74 ± 6 0.14 0.44 
Hekii 60 19 ± 3 57 ± 8 0.43 0.40 
Anakena 65   0.39 0.41 
Anakena 68   0.41 0.43 
Table 6.  Distances and ratios of distances between pukao features and centers of mass.  
Refer to the above diagram and key for clarification of variables.  Note that the center 
of mass lies between 40 and 50 percent up the side of the four pukao considered. 
Physical Form 
Is there any significant spatial variation in the main physical form of pukao?  
Generally, pukao consist of a cylindrical body with an oblong cross section and a top 
cylinder of smaller minimum and maximum diameters.  The maximum body and top 
cylinder diameters in all cases run parallel to the elongation of the base indentation 
(Figure 35).  Members of Cook’s and La Pérouse’s expeditions recorded this general 
form of pukao (Figure 36).  However, some pukao bodies and top cylinders actually 
appear in the form of truncated cones, which will be discussed in the pages that follow.  
I divide my results between those regarding the body of the pukao and those regarding 
its top cylinder. 
 
 
61  
 
Figure 36.  Fragments from engravings based on sketches made by visitors to Rapa Nui 
during Cook’s 1774 visit and La Pérouse’s 1786 visit that highlight pukao.  The upper 
left image is based on a sketch made by William Hodges during Cook’s visit, while the 
other two are based on La Pérouse’s visit (after Heyerdahl and Ferdon, 1961: 54 and 
61, respectively). 
To characterize the cylindrical form of the pukao bodies and top cylinders, I use 
ratio measures, which serves two functions:  (1) it eliminates the aforementioned 
problem of inconsistent embedded scales between pukao models, and (2) it condenses 
observations.  A cylinder with an oblong cross section can be characterized by two 
ratios.  One ratio characterizes how oblong its cross section is (I use minimum diameter 
over maximum diameter), and the other describes how squat it is (I use minimum 
diameter over height). 
In this and subsequent sections, I group observations from pukao at different 
ahu in order to increase sample size in my spatial comparisons.  There are many options 
in choosing spatial units, and the choice of groupings will affect the outcome of any 
subsequent analyses.  However, the division of the pukao-bearing ahu into the three 
 
 
62  
sectors shown in Figure 37 is most reasonable considering that it groups natural clusters 
of ahu and maintains reasonable sample sizes in each sector. 
 
Figure 37.  Sector division of pukao-bearing ahu made for use in spatial comparisons.  
Note that the exact delineations of sectors are arbitrary as long as they maintain the 
existing grouping. 
Body 
Although most pukao are cylindrical, pukao 65 at Anakena (in Sector III) is 
noticeably conical (Figure 38).  Comparison of the minimum and maximum body 
diameters on the tops and bottoms of the 53 scoria bodies outside of Puna Pau reveals 
that a total of six are conical, and these are split evenly between sectors I and III (Table 
7).  Overall, 10% (3 of 30) of the pukao in Sector I have conical bodies, while 30% (3 
 
 
63  
of 10) of the pukao in Sector III have conical bodies.  The observed spatial distribution 
of conical bodies does not significantly differ from expected values (Fisher’s exact test 
in a 2X3 matrix; p = 0.2266). 
The most conical pukao (pukao 65) tapers to approximately two thirds of its 
base diameters toward the top.  Tapering is difficult to discern in the field when the top 
diameter is greater than 80 percent of the bottom diameter.  Also note that, for the three 
pukao in Sector III in which both minimum and maximum diameter tapering can be 
observed, the tapering of the minimum diameter is consistently greater than the tapering 
of the maximum diameter. 
 
Figure 38.  Photograph of the noticeably conical pukao 65 at Anakena (photo by author, 
2014). 
Sector Ahu 
Pukao 
Number 
Top/Bottom 
Max. Diam. 
Percent  
Top/Bottom 
Min. Diam. 
Percent 
I Vinapu 1 8 U 89.22 
I Hanga Poukura 14 85.00 U 
I Hanga Poukura 15 U 92.36 
III Hekii 57 94.22 80.69 
III Anakena 65 67.24 63.10 
III Anakena 68 94.79 91.60 
Table 7.  Tapering in the diameters of pukao with the form of truncated cones.  Lower 
percentages correspond to a greater degree of tapering.  Also note that the indeterminate 
tapering (U) in three cases is due to partial burial and erosion. 
 
 
 
64  
Regarding cylindrical scoria bodies, the pukao with the most circular cross 
section (pukao 29) is located in Ura Uranga te Mahina (Sector I), with a ratio of 
minimum over maximum diameter of 0.9901.  In contrast, the pukao with the most 
oblong cross section (pukao 48) is located in One Makihi (Sector II), with a ratio of 
minimum over maximum diameter of 0.6070.  A comparison of means between sectors 
(Figures 39 & 41) reveals that the mean value in Sector II (n = 7) of 0.695 is 
significantly lower than both the mean value in Sector I (n = 13) of 0.829 and the mean 
value in Sector III (n = 7) of 0.862.  In this and subsequent comparisons, I use a one-
way analysis of variance and the Holm-Sidak method for pairwise multiple comparisons 
in order to identify significant differences between means.  Overall, pukao in Sector II 
have cross sections that are significantly more oblong than those of pukao in the other 
two sectors.  Note that conical pukao (14, 57, 65, and 68) are included in this analysis 
and that diameter ratios at the base of the pukao are considered in all cases.  
The most squat pukao (pukao 38) is in Akahanga (Sector I), with a ratio of 
minimum diameter over height of 1.8815.  The most elongate pukao (pukao 50) is in 
Tongariki (Sector II), with a ratio of minimum diameter over height of 0.8583.  A 
comparison of means between sectors (Figures 40 & 41) reveals that the mean value in 
Sector II (n = 8) of 1.083 is lower than the mean value in Sector III (n = 6) of 1.335 and 
is significantly lower (p = 0.001) than the mean value in Sector I (n = 17) of 1.368.  
Overall, pukao in Sector II are significantly more elongate than those in Sector I.  
Conical pukao are again included in this analysis, and diameter ratios at the base of the 
pukao are considered in all cases. 
 
 
65  
 
Figure 39.  One-way analysis of variance and pairwise multiple comparison applied to 
measures of how oblong pukao body cross sections are between sectors. 
 
 
66  
 
Figure 40.  One-way analysis of variance and pairwise multiple comparison applied to 
measures of how squat pukao bodies are between sectors. 
 
 
67  
Body
I II III
M
in
 D
ia
m
et
er
 / 
M
ax
 D
ia
m
et
er
0.0
0.2
0.4
0.6
0.8
1.0
Top Cylinder
I II III
0.0
0.2
0.4
0.6
0.8
1.0
I II III
M
in
 D
ia
m
et
er
 / 
H
ei
gh
t
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
I II III
0.0
1.0
2.0
3.0
4.0
5.0
6.0
** ***
ns
**
ns
ns
ns
ns ns
*
*ns
Sector
Figure 41.  Spatial comparison of pukao body and top cylinder mean dimensions by sector with 
shape aids for visualization.  Statistical significance is denoted by * (p < 0.05), ** (p < 0.01), 
*** (p < 0.001) and “ns” (p >0.05).  Note in particular the dramatically more elongate top 
cylinders in Sector III. 
Top Cylinder 
Pukao 65 and 67, both at Anakena, do not have top cylinders, but pukao 67 has 
an irregular top stump that may represent the remnant of a broken top cylinder.  With a 
chi-squared test of limited power, we find that top cylinder presence does depend on 
sector (p = 0.0463) but this dependence is not significant when we consider a more 
accurate application of Fisher’s exact test (p = 0.0925).  Although most pukao have 
oblong top cylinders, a few notable exceptions possess truncated inverted oblong top 
cones (Figure 42, Table 8).  Note that all of these exceptions are located in Sector III 
and that maximum diameter tapers more than minimum diameter in all cases.  
 
 
68  
Accordingly, the maximum conical angle of incline is observed when viewing these few 
pukao from the front or back. 
 
Figure 42.  Pukao 66 at Anakena exhibiting an inverted conical protuberance when 
viewed from the front (photo by author, 2014). 
Sector Ahu 
Pukao 
Number 
Bottom/Top Max. 
Diam. Percent  
Bottom/Top Min. 
Diam. Percent 
III Hekii 57 91.45 96.12 
III Anakena 68 88.15 100.00 
Table 8.  Tapering in the inverted conical protuberances of two pukao.  Note that pukao 
66 also clearly has an inverted conical protuberance but that no model was created for 
this pukao. 
Now considering the pukao with cylindrical top protuberances, pukao 18 in 
Vaihu (Sector I) has the top cylinder with the most circular cross section with a ratio of 
minimum over maximum top cylinder diameter of 0.8786.  Pukao 44 in One Makihi 
(Sector II) has the top cylinder with the most oblong cross section with a ratio of 
minimum over maximum top cylinder diameter of 0.4033.  A comparison of means 
between sectors reveals that the mean value in Sector III (n = 3) of 0.636 is lower than 
both the mean value in Sector I (n = 9) of 0.706 and the mean value in Sector III (n = 6) 
of 0.717.  Although the top cylinders in Sector III appear relatively oblong, none of 
these differences are significant.  Note that the inverted truncated cones atop pukao 57 
and 68 are included in this analysis, with ratios taken from measurements at the very top 
 
 
69  
of the pukao.  The relatively oblong top cylinders in Sector III can be explained by the 
small sample size in this sector and the fact that the cross section of the inverted cone 
features becomes more oblong with increasing distance from the pukao body. 
Significant differences exist in the mean squatness values of pukao between 
sectors.  Pukao 51 in Tongariki (Sector II) has the squattest top cylinder with a ratio of 
top cylinder minimum diameter over height of 7.5086.  The most elongate top cylinder 
is found on pukao 68 in Anakena (Sector III), with a ratio of top cylinder minimum 
diameter over height of 0.7073.  A comparison of means between sectors (Figures 41 & 
43-44) reveals that the mean value in Sector III (n = 3) of 1.062 is significantly lower 
than both the mean value in Sector I (n = 14) of 4.222 and the mean value in Sector II (n 
= 6) of 4.470.  Overall, top cylinders in Sector III are significantly more elongate than 
those in Sectors I and II.  Note that the top inverted cones of pukao 57 and 68 are 
included in this analysis and that height in these cases still refers to vertical distance 
between the bottom and top and not the diagonal distance along surface of the cone.  
Comparing squatness between cylinders and inverted cones through the same ratio is 
not problematic in this case for two reasons.  First, Table 8 reveals that minimum 
diameter (involved in the squatness ratio) does not change significantly along the height 
of the truncated cone.  Second, the more elongate top cylinders in Sector III would be 
significantly different from others were we to consider the smaller minimum diameter 
where the truncated cone meets the pukao body. 
 
 
70  
 
Figure 43.  One-way analysis of variance and pairwise multiple comparison applied to 
measures of how oblong top cylinder body cross sections are between sectors. 
 
 
71  
 
Figure 44.  One-way analysis of variance and pairwise multiple comparison applied to 
measures of how squat top cylinders are between sectors. 
Surface Markings 
Side Striations 
Are there any spatial patterns in the presence and types of linear striations found 
on the side of scoria bodies not located in Puna Pau?  The modes within this dimension 
include horizontal striations (H), vertical striations (V), diagonal striations (D), 
combinations, or complete absence of striations (0, Figure 45).  Striations need not 
 
 
72  
extend across the entire side of a scoria body to be considered in this analysis.  Most 
striations are a couple of centimeters deep, and striations rarely extend to the top or 
bottom of scoria bodies.   
   
Figure 45.  Representative samples of a few of the modes within the dimension of side 
striations.  Scoria body 71 from Anakena (left) with horizontal striations, pukao 61 
from Hekii (center) with vertical striations, and pukao 5 from Vinapu with a diagonal 
striation. 
The analysis of surface striations is complicated by the uncertainty involved in 
classifying the side striations of scoria bodies resting on their sides.  Specifically, a 
body resting on its side that has no exposed side striations may in fact have striations on 
its covered surface.  Given this uncertainty, I treat scoria bodies that rest with all sides 
exposed separately from those that do not. 
Of the scoria bodies with all sides exposed, about 62% include side striations 
(Table 9).  Sector I includes the greatest percent of bodies with side striations (71%), 
while Sector III includes the lowest percent with side striations (33%).  However, an 
application of Fisher’s exact test reveals that the spatial distribution of scoria bodies 
with side striations in this sample does not significantly deviate from expected values (p 
= 0.4350). 
Of the scoria bodies with not all sides exposed, a lesser percent (50%) include 
observed side striations, as we would expect (Table 10).  In this case, Sector I includes 
 
 
73  
the greatest percent of scoria bodies with side striations (58%), while Sector II includes 
the lowest percentage of scoria bodies with side striations (25%).  Again, an application 
of Fisher’s exact test reveals that the spatial distribution of bodies with side striations in 
this sample does not significantly deviate from expected values (p = 0.5509).  Overall, 
there is no significant spatial patterning according to sector of scoria bodies with side 
striations and those without. 
Sector Striated Total  Striated (%) 
I 12 17 71 
II 3 6 50 
III 1 3 33 
Total 16 26 62 
Table 9.  Comparison of number of scoria bodies with all sides exposed and striations 
present within each sector relative to total number of scoria bodies with all sides 
exposed within each sector.  Note that an application of Fisher’s exact test reveals that 
the spatial distribution of scoria bodies with side striations in this sample does not 
significantly deviate from expected values (p = 0.4350). 
Sector Striated Total  Striated (%) 
I 7 12 58 
II 1 4 25 
III 3 6 50 
Total 11 22 50 
Table 10.  Comparison of number of scoria bodies without all sides exposed and 
striations present within each sector relative to total number of scoria bodies without all 
sides exposed within each sector.  Note that an application of Fisher’s exact test reveals 
that the spatial distribution of scoria bodies with side striations in this sample does not 
significantly deviate from expected values (p = 0.5509). 
When we compare the abundances of scoria bodies with different striation 
combinations between sectors, we find no significant deviation from expected values.  
This holds true when we consider only scoria bodies with all sides exposed (Table 11) 
and when we consider scoria bodies of all resting faces together (Table 12).  However, 
 
 
74  
note that the p-values from applying the chi-squared test to the data in the tables below 
can be misleading.  This stems from the fact that chi-squared tests become inaccurate 
when multiple expected values are less than one and when over 20% of the expected 
values in the contingency table are less than 5, as is the case in Tables 11 and 12.  In 
such cases, the chi-squared test is less likely to detect differences that actually exist 
between the observation and contingency matrices. 
Sector H V HV VD HVD 0 Total 
I 1 5 1 4 1 5 17 
II 0 2 1 0 0 3 6 
III 0 0 1 0 0 2 3 
Total 1 7 3 4 1 10 26 
Table 11.  Abundances of different types of side striations within each sector for scoria 
bodies with all sides exposed.  Note that a chi-squared test reveals that the abundances 
of different types of side striations in this sample does not significantly deviate from 
expected values (p = 0.711). 
Sector H V HV VD HVD 0 Total 
I 3 8 1 5 2 11 30 
II 0 3 1 0 0 6 10 
III 0 1 2 0 1 6 10 
Total 3 12 4 5 3 23 50 
Table 12.  Abundances of different types of side striations within each sector for scoria 
bodies of all resting faces.  Note that a chi-squared test reveals that the abundances of 
different types of side striations in this sample does not significantly deviate from 
expected values (p = 0.330). 
In order to more accurately identify any potential spatial patterns by sector in the 
abundance of scoria bodies with different surface striation classifications, I conducted a 
series of Fisher exact tests based on the relative abundances by sector of those bodies 
within a given classification mode and those without.  These tests involve 2X3 matrices, 
with sector number against presence/absence of an individual mode, and I again 
consider bodies with all sides exposed separate from the total sample of scoria bodies.  
 
 
75  
Based on Table 13, we find that the existing data suggest that the presence and 
distribution of different types of side striations is independent of sector. 
Striation Type P (sides fully exposed) P (all resting faces) 
V 0.5941 0.6834 
H 1.0000 0.4854 
D 0.3332 0.2055 
HV 0.7271 0.2320 
VD 0.3332 0.3402 
HVD 1.0000 1.0000 
None 0.4350 0.3188 
Table 13.  Results from Fisher tests of individual side striation mode abundances 
between sectors.  Note that more general modes (such as V, H, and D, individually) in 
these tests are taken to include individuals in combination modes (such as HVD).  For 
example, the individual test for vertical striations (V) considers that all scoria bodies 
within modes of V, HV, VD, and HVD belong to the V mode.  The results in this table 
suggest that the distribution of different types of side striations is independent of sector. 
Given the lack of patterning between sectors, we can briefly note the overall 
abundances of each type of side striation (Table 14).  These results indicate that vertical 
striations are by far the most common, and this tendency is accentuated when we 
consider only those scoria bodies with sides fully exposed.  Significant percentages of 
scoria bodies have no striations, while a very limited number include diagonal 
striations.  Indeed, no scoria bodies exist with only diagonal striations. 
 
 
76  
 
Striation Type Presence in bodies with fully 
exposed sides (%) 
Presence in all bodies 
(%) 
V or some combination 58 48 
H or some combination 19 20 
D or some combination 15 10 
None 38 46 
Table 14.  Overall abundances of the three types of side striations when considering all 
scoria bodies and only those with fully exposed sides separately.  Note the relative 
abundance of vertical striations. 
It is possible that the condition of scoria bodies affects the presence of striations.  
In order to estimate the magnitude of this effect, I conducted chi-squared tests of 
observation matrices with condition modes against striation modes.  These chi-squared 
tests have the potential to be inaccurate due to the low values in the contingency matrix 
(previously discussed).  When considering all scoria bodies, the occurrence of various 
striation modes is independent of condition (p = 0.165).  The same holds true to a 
greater extent when we consider only the scoria bodies with all sides exposed (p = 
0.653).  Overall, the observed striation modes are not strongly associated with condition 
modes.  This suggests that striations persist despite decreased condition through erosion 
or that the observed striations are independent of condition because they were applied 
relatively late in the erosive process.  We found that there are no spatial patterns by 
sector in the presence and types of linear striations found on the side of scoria bodies 
not located in Puna Pau.  However, vertical side striations are by far the most common 
type of side striation. 
 
 
77  
Cupules 
Do spatial patterns exist in the presence and forms of cupules found on scoria 
bodies located outside the quarry at Puna Pau?  The modes within this dimension 
include cupules on sides (S), cupules on top (T), cupules on base (B), combinations, 
cupules on unknown face (U), and absence of cupules (0, Figure 46).  Cupules range in 
depth from 2 to 6 centimeters, have clear edges, and are roughly circular in shape. 
   
Figure 46.  Representative samples of a few of the modes within the dimension of 
cupules.  Pukao 73 from Te Pito Kura (left) with side cupules, pukao 10 from Hanga 
Poukura (center) with top cupules, and pukao 14 from Hanga Poukura with a base 
cupule. 
As with the analysis of surface striations, the analysis of cupules on the sides of 
scoria bodies is complicated by the fact that a given body hides the face(s) on which it 
rests.  Specifically, a scoria body resting on its side that has no exposed side cupules 
may in fact have striations on its covered surface.  Because every scoria body has at 
least one hidden face, bias stemming from resting face is inevitable when we compare 
abundances of all cupule modes across space.  This bias can only be eliminated when 
we consider the distribution of individual modes that can be clearly assessed across 
space:  Following this method, we only consider bodies that can definitively be assigned 
membership to a given mode. 
 
 
78  
Of all scoria bodies surveyed outside of Puna Pau, 80% include cupules present 
on at least one face (Table 15).  Sectors II and III include the greatest percentages of 
scoria bodies with side striations (both 90%), while Sector I includes the lowest percent 
of scoria bodies with side striations (77%).  However, the spatial distribution of cupules 
present does not significantly deviate from expected values (Fisher’s exact test; p = 
0.4160). 
Sector With Cupules Total With Cupules (%) 
I 22 30 73 
II 9 10 90 
III 9 10 90 
Total 40 50 80 
Table 15.  Comparison of number of scoria bodies with cupules present within each 
sector relative to total number of scoria bodies within each sector (Fisher’s exact test; p 
= 0.4160). 
When we compare the abundances of scoria bodies with different cupule 
combinations between sectors, we find that there is no significant deviation from 
expected values (Table 16).  Unfortunately, as previously discussed, we cannot compare 
abundances in all modes across space without introducing an unknown degree of bias 
due to resting face.  Also note that, as with the case of side striations, the p-values from 
applying the chi-squared test to the data in the table below can be misleading.  This 
again stems from the fact that chi-squared tests become inaccurate when multiple 
expected values are less than one and when over 20% of the expected values in the 
contingency table are less than 5, as is the case in Table 16. 
 
 
79  
 
Sector S T BS ST BST 0 Total 
I 13 3 1 4 0 8 29 
II 6 1 0 1 0 1 9 
III 7 0 0 0 1 1 9 
Total 26 4 1 5 1 10 47 
Table 16.  Abundances of different types of cupules within each sector for scoria bodies 
of all resting faces.  Three scoria bodies spread evenly among sectors have cupules on 
indeterminate faces and so are not included in this analysis.  The abundances of 
different types of cupules in this sample do not significantly deviate from expected 
values (chi-squared test; p = 0.436). 
In order to more accurately identify any potential spatial patterns by sector in the 
abundance of scoria bodies with different cupule modes, I conducted a series of Fisher 
exact tests based on the relative abundances of those within a given mode and those 
without (Table 17).  Based on these results, the existing data suggest that the presence 
and distribution of different combinations of cupules is independent of sector. 
Cupule Type P (face fully exposed) P (all exposures) 
S 1.0000 0.4062 
T 1.0000 0.7938 
B 0.6703 0.4000 
 
Table 17.  Results from Fisher tests of individual cupule mode abundances between 
sectors.  Limited samples of scoria bodies with a given face under consideration fully 
exposed (center column) are considered separately from the total sample including all 
different types of exposures (right column).  Note that the general modes (such as S, H, 
and D, individually) in these tests are taken to include individuals in combination 
modes (such as BST).  For example, the individual test for side cupules (S) considers 
that all scoria bodies within modes of S, BS, ST, BST belong to the S mode.  
Combination modes were not considered due to their minor abundances.  The results in 
this table suggest that the distribution of different types of side striations is independent 
of sector. 
 
 
 
80  
Given the lack of patterning between sectors, we can briefly turn to the overall 
abundances of each cupule mode.  Cupules on the sides of scoria bodies are by far the 
most common (34 of 50 total, 68%).  A limited number of bodies include top cupules 
(10 of 50 total, 20%), and a very limited number include bottom cupules (2 of 50 total, 
4%).  No scoria bodies exist with only bottom cupules. 
Condition likely affects the presence of cupules.  In order to estimate the 
magnitude of this effect, I conducted a chi-squared test of an observation matrix with 
condition modes against cupule modes (Table 18).  This chi-squared test has the 
potential to be inaccurate given low values in the contingency matrix (previously 
discussed).  When considering all scoria bodies, the occurrence of various cupule modes 
is independent of condition (p = 0.270).  There is a slight tendency of fewer cupules 
present in scoria bodies with lower condition mode.  This may suggest that side cupules 
are diminished through erosion.  However, overall, the observed cupule modes appear 
independent of condition.  Generally, we have found that there are no differences in 
abundances and types of scoria body cupules between sectors.  However, cupules on the 
sides of scoria bodies are the most common. 
 
 
81  
 
Condition S T BS ST BST 0 Total 
P 7 1 1 0 0 5 14 
F 6 3 0 3 1 2 15 
G 12 0 0 1 0 2 15 
E 1 0 0 1 0 1 3 
Total 26 4 1 5 1 10 47 
Table 18.  Observation matrix listing intersections of side cupule and condition modes.  
Note the slight tendency of fewer cupules present in scoria bodies with lower grade 
condition mode.  The values in this observation matrix to not deviate significantly from 
those in the contingency matrix (p = 0.270). 
Puna Pau Scoria Bodies 
To this point, I have reported only on the variation of coastal pukao.  I now take 
some time to discuss the form and surface features of scoria bodies in the quarry at Puna 
Pau.  Most scoria bodies in Puna Pau rest on their sides and are partially buried, thus 
complicating an examination of body dimensions.  However, the scoria bodies of Puna 
Pau include a large amount of surface markings. 
Eleven of the thirteen scoria bodies sampled (about 85%) include side striations, 
and most of these striations are horizontal.  Specifically, eight (about 62%) have 
horizontal striations.  Six (about 46%) have vertical striations, and only one has 
diagonal striations.  Of the bodies with horizontal striations, half include both horizontal 
and vertical striations (about 31% of total).  The majority of Puna Pau scoria bodies 
include side striations, and horizontal side striations are the most abundant. 
Twelve of the thirteen scoria bodies sampled (about 92%) include cupules.  
Given that there is no distinction between top and bottom with the Puna Pau scoria 
bodies, we can only distinguish between cupules on the sides of the cylinder and those 
on the two roughly circular faces.  Seven scoria bodies (about 54%) have cupules on the 
 
 
82  
two circular faces, and seven (about 54%) have side cupules.  Of these, two (about 15% 
of total) include cupules on both the two roughly circular faces and the sides of the 
cylinder.  Overall, the majority of Puna Pau scoria bodies include side cupules, and the 
side cupules are equally distributed by body face. 
Other Sources of Variation 
Additional variation in pukao and other cylindrical scoria bodies stems from the 
presence or absence of the following features: 
• Petroglyphs 
• Slabs Missing 
• Basins 
• Englert Numbers 
• Western Text 
The remaining discussion will focus on each of these dimensions in turn. 
Petroglyphs 
Linton Palmer (1870, 176) was the first visitor to Rapa Nui to report that fallen 
pukao were “all more or less marked by rude carvings of ships, birds, etc.” (Figure 47).  
Heyerdahl and Ferdon (1961, 237) also mention the presence of “incised crescent-
shaped boats” on fallen pukao.  Van Tilburg and Lee (1987, 142) have since 
documented many of the petroglyphs on pukao and other scoria bodies (Figure 48), and 
they hold that many (if not all) petroglyphs were added after pukao fell from moai.  I 
define petroglyphs as any curved lines or collections of lines that do not form linear 
striations.  I also distinguish Western text from petroglyphs.  Most petroglyphs are less 
than a couple of centimeters deep. 
 
 
83  
 
Figure 47.  A watercolor from Linton Palmer’s 1868 visit to the island (top, after Van 
Tilburg, 1992: 171) showing a few of the Puna Pau scoria bodies.  The petroglyphs in 
the scoria body at center in Palmer’s watercolor match those on a scoria body recorded 
by Georgia Lee (1992: 125) and the author (lower right).  Note that this is scoria body 
77 and that a notch was carved into the side of this body sometime between 1868 and 
1992. 
 
 
 
84  
 
Figure 48.  Various tracings of petroglyphs on red scoria pukao, Puna Pau scoria 
bodies, and red scoria ahu lentils, organized by location (after Van Tilburg and Lee, 
1987: 142). 
In their examination of petroglyphs, Van Tilburg and Lee document only those 
located at Puna Pau and five coastal ahu (Figure 48).  I expand on this documentation 
by digitally tracing petroglyphs from pukao at five additional ahu (Hanga Poukura, 
Tarakiu, Ura Uranga te Mahina, Tongariki, and Te Pito Kura).  Petroglyphs are less 
abundant at these five additional ahu than at the five noted by Van Tilburg and Lee, and 
pukao at Tu’u Tahi and Anakena do not include petrogyphs.  Overall, 27 of the 50 
coastal scoria bodies considered (54%) include petroglyphs and 12 of the 13 scoria 
bodies considered in Puna Pau (92%) include petroglyphs.   
Comparing abundances of red scoria body petroglyphs by region, Puna Pau 
scoria bodies include the greatest abundance of petroglyphs (92%), while Sector III 
scoria include the lowest abundance of petroglyphs (40%).  When we consider Puna 
Pau as Sector IV, we find that the spatial distribution of petroglyphs significantly 
deviates from expected values (p = 0.0397, Table 19).  This deviation from expected 
values can be explained by the high abundance of petroglyphs in Sector IV.  Jaussen 
 
 
85  
interprets the abundant petroglyphs in Puna Pau as the “proprietary marks (rona) of the 
owners,” but their age is unclear (Métraux 1940/1970, 303). 
Sector With Petroglyphs Total With Petroglyphs (%) 
I 19 30 63 
II 5 10 50 
III 4 10 40 
IV 12 13 92 
Total 40 63 63 
Table 19.  Comparison of petroglyph presence on scoria bodies by sector.  The spatial 
distribution of scoria bodies with petroglyphs in this sample significantly deviates from 
expected values (Fisher’s exact test; p = 0.0397). 
Taking a closer look at the different types of petroglyphs, I classify them based 
simply on their form in order to avoid ascribing subjective meaning.  Following this 
approach, petroglyphs may be isolated curved lines, collections of lines that create a 
closed form, or circles.  The former is the most abundant, for every scoria body that 
includes petroglyphs includes isolated curved lines (Figure 49).  This allows us to 
conclude that the spatial distribution of isolated curved line petroglyphs significantly 
deviates from expected values (p = 0.0397).  Pointing to Figure 48, Van Tilburg and 
Lee (1987, 143) note the “high degree of conformity between and among the designs at 
all six sites,” but this spatial conformity in general form is evidently not reflected in 
abundances by sector. 
 
 
 
86  
 
Figure 49.  Various pukao with isolated curved line petroglyphs.  (A) Pukao 17 from 
Tarakiu, (B) 19 from Vaihu, (C) 5 from Vinapu, and (D) 18 from Vaihu.  Lee notes that 
“The majority of designs are simple canoe shapes” and speculates that their widespread 
distribution may be because they “marked a victory or conquest over another section of 
the island” (1992, 122/126).  However, note the extensive variability in the form of 
curved line petroglyphs among those shown above. 
Petroglyphs with curved lines that define closed forms also occur on scoria 
bodies, but they are restricted to Sectors I and IV (Table 20, Figure 50).  This mode of 
petroglyph is most abundant in Puna Pau (about 62%) and the spatial distribution of this 
petroglyph mode deviates quite significantly from expected values (p = 0.00038). 
 
 
87  
 
 
 
 
 
 
Table 20.  Comparison of closed curved form petroglyph presence on scoria bodies by sector.  
The spatial distribution of scoria bodies with closed curved form petroglyphs in this sample 
significantly deviates from expected values (Fisher’s exact test; p = 0.00038). 
Sector 
With Closed 
Curved Forms Total 
With Closed 
Curved Forms (%) 
I 4 30 13 
II 0 10 0 
III 0 10 0. 
IV 8 13 61.54 
Total 12 63 19.05 
 
 
88  
 
Figure 50.  Various scoria bodies with closed curved form petroglyphs.  (A) scoria body 
80 from Puna Pau, (B) pukao 35 from Akahanga, (C) pukao 36 from Akahanga (after 
Van Tilburg and Lee, 1987: 145), and (D) pukao 36 from Akahanga.  Van Tilburg and 
Lee refer to the forms from Akahanga as “birdman symbols” (1987, 143), but Lee later 
refers to that at bottom as a two-headed frigate bird (1992, 20). 
 
 
89  
Just like closed curved form petroglyphs, circle petroglyphs are restricted to 
Sectors I and IV (Figure 51).  Three pukao in sector I include circles, and one scoria 
body in Puna Pau includes a circle.  Based on Fisher’s exact test the spatial distribution 
of this petroglyph mode does not deviate significantly from expected values (p = 
0.8101). 
 
Figure 51.  Pukao 4 from Vinapu with isolated curved lines, a cupule, and circles at the 
lower left. 
Finally, in four pukao, relatively large pairs of side cupules located near the base 
that are horizontally level may be interpreted as a fourth type of petroglyph.  Van 
Tilburg (1992, 200) specifically refers to pukao 41 at Tu’u Tahi (Figure 52), which she 
claims has been “recarved with Make Make eyes.”  According to Heyerdahl and Ferdon 
(1961, 41), Make Make is “the principal Easter Island god.”  Distinctly carved pairs of 
cupules such as those on the pukao at Tu’u Tahi also appear on three other pukao at 
different locations (Figure 53). 
 
 
90  
 
Figure 52.  Drawing by P. Cristián Arévalo of pukao 41 at Tu’u Tahi with horizontally 
aligned side cupules (after Van Tilburg, 1992: 200). 
 
Figure 53.  The four pukao that include horizontally-aligned pairs of side cupules near 
their bases.  (A) Pukao 41 from Tu’u Tahi, (B) 23 from Vaihu, (C) 51 from Tongariki, 
and (D) 57 from Hekii. 
 
 
 
 
91  
Slabs Missing 
Along the Rapa Nui coastline, there exist five pukao with slabs of scoria 
removed along the upper edge of the body (Figure 54).  At least some of these features 
were present during Cook’s visit to the island in 1774, for he notes that “In some, the 
upper corner of the cylinder was taken off in sort of a concave quarter-round; but in 
others the cylinder was entire” (1777/2005, 160).  During his 1868 visit to the island, 
Palmer (1870, 179) also notes notches in a scoria body at Akahanga: “the top was flat, 
and cut away on each side, so as to make a step or shelf.”  This description matches the 
current appearance of pukao 34 at Akahanga.  Regarding this pukao, Palmer (1870, 
179) continues: “On it I found two skulls, very much perished, which, from the 
dentition, I judged to be those of youths of twelve or fourteen years old.”  Métraux 
(1940/71, 302) later notes Palmer’s association between death and the pukao with slabs 
missing and asserts that such pukao “seem to have been supports for a wooden frame on 
which corpses were exposed.”  Katherine Routledge (1919/1998, 229) holds a similar 
view:  There exist pieces of pukao and moai at an ahu along the southern coast that 
were “specially used for exposing ‘fish men,’” otherwise referred to as “corpses of slain 
men.”  Historic sources suggest that the relatively few fallen pukao with slabs missing 
served a purpose in funerary rituals, but it is also possible that people removed such 
slabs for other purposes (e.g. building materials or ahu lentils). 
 
 
 
92  
 
Figure 54.  Four of the pukao that have large slabs missing from the upper body edges.  
(A) Pukao 14 from Hanga Poukura, (B) 34 from Akahanga, (C) 38 from Akahanga, and 
(D) 49 from Tongariki (photos by author, 2014).  The fifth pukao with a slab missing is 
located at Hekii. 
Several of the Puna Pau scoria bodies also have slabs missing, but these slabs 
tend to be more irregular in form.  For example, recall the rectangular notch carved into 
scoria body 77 following the creation of Palmer’s watercolor depiction (Figure 47).  
Scoria body 80 at Puna Pau includes a notch similar in size, shape, and placement, and 
scoria body 86 (Figure 55) has a large slab missing from its base, which leaves a 
significant concave hollow beneath it.  This hollow was quite likely present during 
Métraux’s visit to the island:  During a rainy day at Puna Pau, he “took refuge inside 
one of these cylinders that had been converted into a shelter” (Métraux 1957, 201).  
Given that this scoria body did not have this concave hollow during Palmer’s visit, we 
 
 
93  
can date the removal of this slab to sometime between 1868 and 1934.  A few irregular 
slabs are missing from Puna Pau scoria bodies for unknown reasons. 
 
Figure 55.  Scoria body 86 from Puna Pau in upper photo at right with concave hollow 
missing (photo by author, 2014).  In Linton Palmer’s 1868 watercolor (bottom, after 
Van Tilburg, 1992: 171), these three scoria bodies are viewed from the opposite 
direction, so the scoria body with the current hollow is at left in the watercolor.  
Overall, the current concave hollow was created sometime between 1868 and 1934. 
Basins 
Two pukao located along the coastline have large carved basins with an oblong 
cross section (Figure 56).  Heyerdahl and Ferdon note the basin in the pukao at Vinapu 
 
 
94  
and conducted a shallow excavation surrounding it (Figure 57).  They it to be a “topknot 
resting upside down” (Heyerdahl and Ferdon 1961, 119), as do I, but did not excavate 
beneath it for verification.  When Heyerdahl and Ferdon (1961, 119) observed this 
scoria body, “The concavity was partially filled with earth, which we removed to 
disclose two human deciduous molar teeth lying on its bottom.”  They also note that 
“About two-thirds of the distance from the seaward end was found an elongated ovoid 
cobble which Martín Rapu thought may have been a head rest for a cadaver” 
(Heyerdahl and Ferdon 1961, 119).  These observations lead them to speculate that 
“The concavity may have been used as a form of drying platform” (Heyerdahl and 
Ferdon 1961, 119).  However, it is just as likely that the two red scoria bodies with 
basins carved were used to collect rainwater.  Métraux (1957, 66) notes the use of 
carved stone basins called taheta:  These “holes cut artificially in the rocks were – if the 
natives of our own day are to be believed – little reservoirs that served in days gone by 
to catch rainwater.” 
  \  
Figure 56.  The two pukao located along the southern coast with basins dug: Pukao 4 
from Vinapu (top) and pukao 48 from One Makihi (bottom, photos by author, 2014). 
 
 
 
95  
 
Figure 57.  Shallow excavation conducted by the Norwegian Archaeological Expedition 
in 1955 around the base of pukao 4 (after Heyerdahl and Ferdon, 1961: Appendix). 
In addition to the two bodies with oblong basins, there are three with notable 
circular depressions of lesser volume in the center of the side of the pukao body that 
currently faces upward (Figure 58).  Georg Forster noted a scoria body during his 1774 
visit to the island at a poorly defined location that “had a hole on each side, as if it had 
been made round by turning” (Kahn 1968, 325).  No subsequent visitor makes note of 
such holes.  I witnessed no bodies with holes on both sides, but the three in Figure 58 
may have holes on the face on which the body currently rests. 
 
 
96  
 
Figure 58.  Top views of three pukao with notable circular holes in the center of the 
upward face.  Pukao 14 is from Hanga Poukura (left), pukao 17 is from Tarakiu, and 
the pukao 34 is from Akahanga.  The former pukao rests on its top, while the latter two 
rest on their bases. 
Historic sources reveal a final past embellishment of pukao that appears to have 
left no remaining trace: Piles of white stones atop pukao.  These piles may or may not 
be related to the presence of cupules on the tops of some pukao.  Based on his 1722 
visit to the island, Jacob Roggeveen provides the first reference to bodies of some white 
material resting atop moai: “They have on the head a basket heaped up with flints 
painted white deposited in it” (Ruiz-Tagle 2005, 32).  Given that the pukao that we see 
today are not baskets, the editor notes that “Roggeveen’s description of the statues 
seems to show that he never got close to one of them, but saw them only from a 
distance of some hundreds of yards” (Ruiz-Tagle 2005, 32).  However, it is possible 
that both organic baskets and red scoria pukao previously adorned moai.  Also note that 
the white pebbles that Roggeveen notes are not entirely fictional, for Spanish visitors 
noted something similar forty eight years after Roggeveen’s visit.  Specifically, Felipe 
González de Haedo notes that “there is a small concavity on the upper surface of the 
[head-dress] in which they place the bones of their dead, from which it may be inferred 
that they serve at once for idols and funeral pyres” (Ruiz-Tagle 2005, 57).  However, 
Métraux (1940/1971, 300) and Heyerdahl and Ferdon (1961, 110) believe that the 
 
 
97  
Spanish mistook coral fragments atop pukao for bone.  In excavations of Vinapu, the 
Norwegian Archaeological Expedition came across “many small, smooth, beach 
pebbles of coral which varied from about 0.5 to 15 cm. in diameter.  Juan Chavez said 
that he had heard that in earlier times people threw these on the tops of the topknots and 
that many were nearly covered with them” (Heyerdahl and Ferdon 1961, 110).  The 
only other reference to white stones is one that Routledge (1919/1998, 199) briefly 
makes: “[The pukao] are said by the present inhabitants to have been kept in place by 
being wedged with white stones.”  Overall, white stones involved in pukao adornment 
offer tantalizing glimpses into the past but evidently left very little trace on the physical 
form of the existing pukao with the possible exception of a couple of the holes noted in 
Figure 58. 
Englert Numbers 
Painted numbers constitute some of the most recently added variation among 
scoria bodies (Figure 59).  Father Sebastian Englert, a Capuchin Franciscan friar who 
lived on Rapa Nui from 1935 until his death in 1969, applied these numbers to scoria 
bodies, moai, and various other archaeological features spread across the island.  
Though an eyesore, these numbers provide a diagnostic feature for scoria body 
identification.  Of the scoria bodies considered, a total of 27 of 66 (about 41%) include 
Englert numbers.  Of these, 22 of the 53 bodies observed along the coast (about 42%) 
and 5 of the 13 scoria bodies observed in Puna Pau (about 38%) include Englert 
numbers. 
 
 
98  
   
Figure 59.  Two scoria bodies with clearly-marked Englert numbers:  Pukao 65 from 
Anakena (left) and scoria body 88 from Puna Pau (right, photos by author, 2014). 
Western Text Engravings 
Aside from Englert numbers, a few scoria bodies include engraved Western text 
(Figure 60).  Western text engravings are most abundant in Puna Pau, for 5 of the 13 
Puna Pau scoria bodies (about 38%) include such engravings, while only two pukao 
along the coastline (one at Vinapu and the other at Akahanga) include such engravings.  
The origin of these engravings is unknown, but it is understandable to find the most 
abundant graffiti in Puna Pau given its close proximity to Hanga Roa, which is the only 
current residential area on Rapa Nui. 
    
Figure 60.  Two scoria bodies at Puna Pau (77 at left and 80 at right) with engraved 
Western text. 
 
 
99  
Conclusion 
Implications of Pukao Dimensions for Transport 
Volume 
Though insignificant (r2 = 0.344), the slight positive correlation between pukao 
volume and distance from quarry (Figure 31) indicates that pukao distant to the quarry 
represent greater investments of energy than those located nearer to the quarry.  A 
greater number of volume measurements of higher precision are needed in order to 
better test the strength of this correlation.  If this correlation remains significant, then it 
may be explained by future research.  It may be the case that the Rapanui simply moved 
larger scoria bodies further across the island in order to adorn the heads of possibly 
larger moai that are located closer to the main statue quarry (recall that Puna Pau and 
Rano Raraku are located on nearly opposite ends of Rapa Nui, Figure 1). 
Base Indentation 
The presence of base indentations on all well-preserved pukao with bases 
exposed and the associated scarcity of base striations have implications regarding 
transport.  This set of observations suggests that the Rapanui did not slide pukao onto 
the top of moai for three reasons.  First, it is difficult to carve a properly-oriented base 
indentation to match the top dimensions of a moai when the pukao is resting on its base 
prior to sliding.  Second, shallow base indentations made in the relatively friable red 
scoria of the pukao would likely be erased by friction associated with sliding a pukao 
over much distance.  Finally, if the Rapanui slid pukao into place, then we would expect 
some preserved wear deformation in the base of pukao from friction (refer to Table 3).  
 
 
100  
No amount of refinishing could erase scratches due to sliding on the base once the 
pukao is placed atop the moai and the face between the two becomes sealed.  The mere 
presence of the base indentation free of striations negates the hypotheses involving 
sliding:  including the Tower Hypothesis (attributed to Vincent Lee, Figure 11) and the 
Wooden Ramp Hypothesis (attributed to Pavel Pavel, Figure 13). 
Aside from not explaining observations from the archaeological record, the 
Wooden Ramp Hypothesis loses its physical feasibility when applied to large pukao 
masses and moai heights.  Pavel Pavel’s (1995, 71) experiment involved raising a 900 
kilogram pukao to the top of a 3 meter moai using two parallel beams of wood.  Though 
hardwood from the relatively small Sophora toromiro tree was available in the past on 
Rapa Nui (Horrocks et al. 2012), the wood involved in the Wooden Ramp Hypothesis 
most likely would have been collected from the now-extinct palm Pashalococos 
disperta, which was likely closely related to Jubaea chilensis (Gurley and Liller 1997, 
82).  Palm trunks are a poor material for bearing loads because palms are actually large 
grasses with spongy interiors.  Jubea chilensis reaches an average height of 20 meters 
after 60 years of growth, and, on its side, the trunk can support about 6 tons (Gurley and 
Liller 1997, 83).  Despite this strength estimate, relatively massive pukao (upwards of 
10 metric tons) applying a torque with a lever arm of up to 10 meters would very likely 
cause a wooden ramp made of two unsupported trunks of Jubea chilensis to yield.  
Physical Form 
Despite the fact that most pukao share a cylindrical form with an oblong 
horizontal cross section, the physical form of pukao does not appear to hold immediate 
implications regarding transport.  For example, the oblong section shared by all pukao 
 
 
101  
potentially reflects an aspect of the transport method that involved rolling up a ramp.  
Rolling can be achieved with oblong cross sections, but it is important to note that 
oblong sections are not incompatible with any other proposed transport methods.  While 
the presence of pukao with conical forms (e.g. Figure 38) argues against any hypothesis 
involving rolling up a ramp, it is possible that prehistoric Rapanui carved pukao to their 
current form only after placing the pukao atop moai.  The possibility of finishing pukao 
carving at the top of moai may be supported by the fact that the best preserved pukao at 
Anakena have exceptionally smooth sides with few to no surface markings:  I assume 
that these smooth pukao surfaces were not refinished during the recent reconstruction of 
Ahu Nau Nau at Anakena. 
Carved top cylinders display a high degree of variability throughout the island, 
with significant differences between the northern and southern coasts (Figure 41).  If the 
presence of top cylinders holds any functional value, then it must be a value that leaves 
top cylinder form quite unconstrained. 
Surface Variation 
The surface variation of pukao may hold implications regarding transport, but 
they are difficult to establish when we consider that (a) the Rapanui may have 
refinished pukao after they reached the top of moai and (b) the existing surface variation 
may be a product of activity following the toppling of pukao.  Abundant horizontal 
striations found along the side of scoria body 71 near Anakena (Figures 9, 30, and 45) 
can be interpreted as scratches associated with sliding while rolling.  The abundance of 
horizontal striations on many of the Puna Pau scoria bodies can be interpreted similarly.  
This observation and interpretation aligns with the established reasoning that the 
 
 
102  
Rapanui rolled relatively large bodies of scoria across the island for use at coastal ahu.  
It is important to note that the absence of abundant horizontal striations on the sides of 
pukao does not rule out a transport hypothesis involving rolling up a ramp. First, sliding 
may not have taken place while rolling a pukao up a ramp.  Second, striations from 
sliding up a ramp may have been erased when the Rapanui refinished pukao at the top 
of a ramp. 
The horizontal and abundant vertical striations found on pukao may support 
William Mulloy’s Simultaneous Set Up Hypothesis (1970, 16, Figure 12).  With this 
interpretation, striations represent notches used to help affix pukao to moai during the 
raising of the moai.  However, few pukao include both horizontal and vertical striations. 
Vertical striations are rarely wide enough to act as notches for bound beams (although 
they are wide enough for rope), and they rarely extend along an entire face of the pukao. 
The possibility of functional value associated with side cupules is similarly 
inconclusive.  Cupules most often found on the curved sides of pukao may have aided 
in gaining traction while rolling pukao up a ramp, but side cupules may again represent 
late additions that came following the toppling of pukao.  For example, Lee (1992, 124) 
speculates that cupule carving in pukao and moai is “a case of trying to extract the mana 
from them, or to release, and thus destroy, the mana inherent in them.”  Better 
information on the timing of modifications to pukao is required to discern the sequence 
of events inherent in their manufacture. 
The Modified Ramp Hypothesis 
Upon first glance, most of the variable pukao dimensions appear to have 
inconclusive associations with transport function.  Only the presence of base 
 
 
103  
indentations helps us confidently narrow down the hypotheses to either those involving 
rolling up a ramp or simultaneous moai and pukao set up.  As mentioned earlier, Lipo et 
al. (2013) used archaeological observations and experimental verification to firmly 
establish ‘walking’ as the prehistoric means of moai transport.  This work allows us to 
make further inference regarding pukao transport by significantly diminishing the 
viability of Mulloy’s hypothesis involving simultaneous set up of moai and pukao from 
a horizontal position.  Specifically, this leaves us with some form of the Ramp 
Hypothesis for the majority of pukao transport.  What follows is an exploration of the 
physical feasibility of the Ramp Hypothesis and a summary of how a modified version 
of this hypothesis (the Modified Ramp Hypothesis) successfully explains observations 
of the archaeological record.  In this exploration, I make assumptions that establish the 
minimum physical requirements for the Modified Ramp Hypothesis. 
The first question that arises when considering any version of the Ramp 
Hypothesis is which direction does the ramp extend.  Given that seawalls of ahu often 
drop precipitously to the adjacent intertidal zone and that moai are aligned along the 
length of ahu parallel to the coastline, it would be most feasible to construct a ramp for 
pukao transport on the inland side of the ahu.  Similarly, I assume that the Rapanui used 
a straight ramp coming off of the front of moai instead of a zig-zag situation stacked in 
front of a moai, for the latter situation limits the number of people that can be involved 
in applying a force for transport. 
I consider two possible techniques of rolling a pukao up a ramp: pulling from 
above and/or levering from below.  Pulling from above can be accomplished by using 
rope in a parbuckle setup (Figure 61).  In a parbuckle setup, one passes a doubled rope 
 
 
104  
around a cylindrical object on its side, anchors one end of the doubled rope at a point 
up-ramp of the object, and pulls on the other end of the doubled rope.  Note that either 
stakes at the top of the ramp or the top of the moai can serve as the anchor in the 
parbuckle setup.  Rope was certainly available to the Rapanui, for ethnohistoric 
accounts document the local tradition of producing cordage and rope (hau) from the 
bark of Triumfetta semitriloba, a woody shrub that still grows on the island (Métraux 
1940/1970, 210).  Additionally, levering a pukao from below can be accomplished by 
using relatively small branches from trees, such as Sophora toromiro, that previously 
grew on the island (Horrocks et al. 2012). 
 
 
105  
 
Figure 61.  The configuration of a parbuckle.  Note the parbuckle anchor behind the 
person and that such parbuckles were once commonly used in preindustrial Europe for 
transporting barrels up and down inclines.  Source: “Parbuckle – Chambers’ Twentieth 
Century Dictionary of the English Language.” Wikipedia: The Free Encyclopedia. 
Wikipedia Foundation, Inc. 17 Jan. 2012. Web. 20 May 2015. 
Physics Derivation: Pushing through Levering   
We would like to find a simplified equation that relates the angle of incline of 
the ramp to the force due to people applied to the upper portion of a lever in contact 
with a pukao, as shown in Figure 62. 
 
 
 
106  
 
Figure 62.  Forces involved in levering a pukao of radius R up a ramp, where  gives 
the angle of incline of the ramp.  The force on the pukao due to gravity acts downward 
at the center of mass of the pukao and is given by mg, where m is the mass of the pukao 
and g is the acceleration due to gravity.  The normal force due to the ramp acts 
perpendicular to the ramp and is designated by .  A person exerts a force  on the 
top of the lever, and this force is communicated through force  where the lever and 
pukao touch.  The angle  here represents the angle formed by a line normal to the 
ramp surface and a line normal to the lever.  Note that we know that the lever case of 
static equilibrium is stable only if .  The reference for this situation is modified 
for convenience such that x is parallel to the inclined ramp surface. 
We begin by considering the sum of forces in the x- and y-directions when the 
pukao is in a state of static equilibrium.  Note that the x and y components of the force 
acting between the lever and pukao at the point of contact are given by   and , 
respectively. 
 
  (1) 
 
  (2) 
 
 
107  
Given how we previously defined  and  , we can relate   and  to  
through the following two equations. 
 
  (3) 
 
  (4) 
 
By substituting equation (3) into equation (1), we come to equation (5), and by 
substituting equation (4) into equation (2), we come to equation (6), as seen below. 
 
  (5) 
 
  (6) 
 
Given that the normal force is of little interest to us, we will continue by using 
only equation (5).  In order to find  in terms of , we must now consider the sum of 
torques acting on the lever, as outlined below in Figure 63. 
 
 
 
108  
 
Figure 63.  Torques acting on the lever as it pivots around point P.  Here, the force  
pushing back from the pukao acts at a distance d from the pivot point.  Meanwhile, the 
force  acts in the positive direction at a distance k from the pivot point. 
 
Given that the situation in Figure 63 is in static equilibrium, we can express the 
torques in the system through equation (7). 
 
  (7) 
  
Solving equation (7) for , we come to equation (8). 
 
 
 
(8) 
   
Substituting equation (8) into equation (5), we come to equation (9). 
 
 
 
(9) 
 
Given equation (9), we would now like to get d and  in terms of known 
variables.  Starting with d, consider Figure 64 below. 
 
 
109  
 
 
Figure 64.  An analysis of the distance d and how it relates to .  Two lines tangent to a 
circle intersect at a distance d from their individual points of tangency.  Because the 
radius of the circle (R) is constant, two triangles can be drawn with a shared 
hypotenuse.  Thus, we have two identical triangles and know that their hypotenuse must 
bisect the angle  as shown. 
 
By examining one of the triangles formed Figure 64, we find equation (10). 
 
 
 
(10) 
 
Solving equation (10) for d, we come to equation (11). 
 
 
 
(11) 
 
 
 
110  
 
We now substitute equation (11) for d into the leftmost term in equation (9) and 
come to equation (12) by expanding the tangent term. 
 
 
 
 (12) 
  
 
Equation (12) can be rewritten as follows using a trigonometric identity for sine. 
 
 
 
The above term simplifies to the following. 
 
 
 
Adding the above modified term back into equation (9), we come to equation (13). 
 
 
 
 
 (13) 
  
 
We find the equation (14) below by solving equation (13) for . 
 
 
 
 
 (14) 
  
 
 
 
111  
 
We now consider the upper limit of  when , as shown below in 
equation (15).  This situation corresponds to an overall vertical lever force that acts 
directly against the full force due to gravity of the pukao. 
 
 
 
 
 (15) 
  
 
The right term in equation (15) can be expanded to the following using a 
trigonometric identity for sine. 
 
 
 
By simplifying the above term in equation (15), we come to equation (16). 
 
 
 
(16) 
 
In this equation, the variables m and R are fixed for a given pukao, the length of 
the lever k can be approximated by considering a reasonable height of a person.  Thus, 
we have found an expression that relates the angle of incline of a ramp  to the 
maximum force that people exert on the pukao through a lever. 
Physics Derivation:  Pulling with Rope 
We would like to find a simplified equation that relates the angle of incline of 
the ramp to the force due to people applied to the top of the pukao through pulling with 
rope through parbuckling, as depicted below in Figure 65.  In this situation, we assume 
that the applied force is roughly parallel to the ramp surface.  The latter assumption is 
 
 
112  
reasonable given the height of pukao on its side, the length of the ramp, and the height 
of a person.   
 
 
Figure 65. Forces involved in pulling a pukao of radius R up a ramp, where  gives the 
angle of incline of the ramp.  The force on the pukao due to gravity acts downward at 
the center of mass of the pukao and is given by mg, where m is the mass of the pukao 
and g is the acceleration due to gravity.  The normal force due to the ramp acts 
perpendicular to the ramp and is designated by .  A person exerts a force  on the 
rope, and this force is communicated to the top of the pukao.  A force of friction 
designated by  acts as shown in order to prevent the pukao from slipping.  The 
pukao pivots as it rolls around the point P.  The reference for this situation is modified 
for convenience such that x is parallel to the inclined ramp surface. 
 
Given that the situation in Figure 65 is in static equilibrium, we can chose to 
examine both the sum of forces acting in the x- and y-directions and the sum of torques.  
Given that the friction force is difficult to work with, we eliminate it by considering the 
 
 
113  
sum of torques acting on the pukao around the pivot point P.  This leads us to equation 
(17). 
 
  (17) 
 
Solving equation (17) in terms of , we come to equation (18). 
 
 
 
(18) 
 
In equation 18, we have found an equation that relates the angle of incline of the 
ramp to the force due to people applied to the top of the pukao through pulling with 
rope.  In this relatively simple situation, the applied force needed to move the pukao is 
only dependent on the mass of the pukao and the angle of incline. 
Implications of Transport Equations 
The derived transport equations (summarized in Figure 66) help describe pukao 
transport.  Below, I explore the implications of these equations for pukao transport by 
considering measurements of pukao, moai, and ahu at four different locations on Rapa 
Nui (Table 21).  Generally, Newtonian physics and human strength estimates verify that 
pukao transport by rolling up a ramp is physically feasible. 
 
 
 
114  
 
Figure 66.  Equation giving the force required to move a pukao up a ramp by pushing 
with a lever (left, equation 16) or by pulling with a parbuckle (right, equation 18).  In 
both equations, the F term gives the force required to move the pukao, m gives the mass 
of the pukao, g stands for the acceleration due to gravity, and  gives the ramp angle of 
incline.  In the pushing/levering equation at left, R refers to the radius of the pukao, and 
k refers to the length of the lever. 
ahu  h (m) Pukao Number m (kg) R (m) Pukao h (m) k (m) Vaihu 8 23 6,383 ± 2,920 1.2 1.2 1.4 Ura Uranga te Mahina 8 29 3,541 ±726 1.0 1.1 1.4 Akahanga 8 38 4,170 ± 603 0.7 1.0 1.4 Te Pito Kura 10 73 11,500 1.3 1.8 1.4 
Table 21.  Estimated absolute scale values for dimensions of pukao, moai, and ahu at 
four locations used in sample physical calculations graphs.  True to the equations 
derived in the Appendix, h denotes the combined height of the erect moai and ahu over 
the surrounding ground, m denotes the mass of the pukao, R denotes the radius of the 
pukao, and k denotes the estimated lever arm associated with pushing.  An estimate of 
1.4 meters for k is reasonable when assuming that human height ranges between 1.5 and 
1.6 meters.  Measurements from Te Pito Kura are taken from Heyerdahl and Ferdon, 
1961.  Other measurements come from those taken in the field by teams under Hunt and 
Lipo, but pukao mass estimates come from the models that I generated in this project. 
Graphing the force required to pull or lever pukao up a ramp with variable angle 
of incline at each location by considering the equations in Figure 66 and the values in 
Table 21 produces Figure 67.  This figure, combined with basic human strength 
estimates, supports the conclusion that relatively small numbers of people can transport 
a pukao up a ramp with angles of incline between 5 and 20 degrees.  In this assessment, 
 
 
115  
I rely on Snook and Ciriello’s (1991) comprehensive tables of human strength 
measurements that are frequently used in occupational safety planning.  Though 
conservative (strength estimates are intended to avoid long term lower back pain), 
estimates for pushing and pulling assume that some portion of the population should be 
capable of pulling or pushing a mass of about 70 kg (about 686 N). 
 
Figure 67.  Applied force required to move pukao up ramps with variable angles of 
incline at four different ahu (1 = Te Pito Kura, 2 = Vaihu, 3 =  Akahanga, and 4 = Ura 
Uranga te Mahina).  Solid black lines represent force required with only pulling, and 
dashed lines represent force required with only pushing.  When we consider a human 
strength estimate of 70 kg (686 N) per person, we can draw the three horizontal red 
lines above, which approximate the forces that can be applied by 5, 10, and 15 
individuals. 
Generally, the largest pukao (1, that at Te Pito Kura) requires the greatest force 
to roll at any given angle of incline.  The pukao at Vaihu (2) requires an intermediate 
applied force, while those at Akahanga (3) and Ura Uranga te Mahina (4) require the 
 
 
116  
lowest applied force.  Note that pushing through levering becomes more efficient than 
pulling when the radius and mass of the pukao are sufficiently low.  In the case of 
Akahanga, the relatively small radius of the pukao enables it to be levered with a lower 
applied force than the less massive pukao at Ura Uranga te Mahina.  However, only a 
limited number of people can lever behind a pukao (due to crowding), while many more 
can parbuckle from above.  Indeed, an advantage of pulling up a straight ramp is that it 
maximizes the number of people that can apply a force.  In Figure 67, the x-coordinate 
of the intersection of the red and black lines give estimates of angles of incline possible 
with different numbers of people at different locations.  For example, at Vaihu (2), 
pulling groups ranging in size between 5 and 10 individuals are capable of pulling the 
pukao up a ramp of angle of incline between about 6 and 13 degrees. 
William Mulloy acknowledges the possibility of such transport but objects that 
ramps (specifically at Te Pito Kura) would need to be “several hundred metres long – a 
seemingly prohibitive amount of construction” (1970, 15).  However, we find that 
ramps several hundred meters long are not needed at any location when we plot force 
required to move a pukao against the length of the ramp (Figure 68).  Even at Te Pito 
Kura, just ten people pulling the pukao can do so with a ramp length of less than one 
hundred meters (Figure 68). 
 
 
117  
 
Figure 68.  Applied force required to pull pukao up ramps with variable angles of 
incline at four different numbered ahu (1 = Te Pito Kura, 2 = Vaihu, 3 =  Akahanga, 
and 4 = Ura Uranga te Mahina).  Solid lines represent force required with only pulling.  
When we consider a human strength estimate of 70 kg (686 N) per person, we can draw 
the three horizontal red lines above, which approximate the forces that can be applied 
by 5, 10, and 15 individuals. 
Figure 68 shows that the largest pukao (1, that at Te Pito Kura) requires the 
longest ramp for any given applied force, followed by that at Vaihu (2), Akahanga (3), 
and Ura Uranga te Mahina (4).  As in Figure 67, when we consider a human strength 
estimate of 70 kg (686 N) per person, we can draw the three horizontal red lines above, 
which approximate the forces that can be applied by 5, 10, and 15 individuals.  Thus, 
the x-coordinate of the intersection of the red and black lines again give estimates of 
 
 
118  
ramp lengths possible with different numbers of people at different locations.  For 
example, at Vaihu (2), groups ranging in size between 5 and 10 individuals are capable 
of pulling the pukao up a ramp with a length between about 23 and 32 meters. 
Thus far, we have established that relatively small numbers of people can move 
pukao up ramps of reasonable dimensions by parbuckling and/or levering.  Another 
important consideration regarding the feasibility of the Modified Ramp Hypothesis is 
whether or not the volumes of such ramps are reasonable.  Ramps would likely be made 
of loosely consolidated stone and soil.  I consider ramp volumes to be reasonable if they 
are less than the approximate volumes of corresponding ahu. 
Physics Derivation: Ramp Volume Estimates 
We wish to roughly examine the volume of material that composes a ramp with 
an angle of incline of , as depicted below in Figure 69. 
 
Figure 69.  A ramp with an angle of incline of , height h, width w, and length a.  In 
this case, h is fixed as the height of the moai an ahu, w can be approximated by the 
height of the pukao (its width when rolled on its side) plus some buffer room, and  
(with a corresponding a) can be chosen. 
 
 
 
119  
We know that the volume (V) of the above ramp is given by the following equation 
(19). 
 
 
 
(19) 
 
We also know that the length of the ramp can be given in terms of the height and 
angle of incline as shown in equation (20). 
 
 
 
(20) 
 
By substituting equation (20) into equation (19), we come to our result of 
equation (21). 
 
 
 
(21) 
 
Note that we can improve on the volume estimate given in equation (21) by 
considering the additional volume that comes from material sloping off the sides of the 
ramp at a given angle of repose.  In this situation, each side slope approximately forms a 
tetrahedral shape.  The volume for a tetrahedron ( ) is given by equation (22) below, 
where  represents the area of its base and h represents its height.  Note that h in this 
case is equivalent to the h in equation (21). 
 
 
 
(22) 
 
We wish to find  in terms of the height h and a given angle of repose .  
Without considering the angle of repose, we have equation (23), where x gives the 
width of the tetrahedron and a is the same as in Figure 69. 
 
 
120  
 
 
 
(23) 
 
We can relate x to the known variables of  and h through the following equation (24). 
 
 
 
(24) 
 
Substituting equation (24) into equation (23), we come to equation (25) below. 
 
 
 
(25) 
 
Note that we can find equation (26) below by substituting equation (20) into 
equation (25). 
 
 
 
(26) 
 
Substituting equation (26) into equation (22), we come to our final result in 
equation (27), which gives the volume of a tetrahedron in the desired terms. 
 
 
 
(27) 
 
Given that one such tetrahedron comes of each of the two flanks of the ramp, we 
combine two of the volumes given by equation (27) with the volume for the main body 
of the ramp given by equation (21) and come to our new result in equation (28). 
 
 
 
(28) 
 
Equation (28) gives a fairly accurate ramp volume estimate in which we are free 
to choose a reasonable ramp path width (w), ramp angle of incline , ramp height (h), 
and side ramp angle of repose ( ). 
 
 
 
121  
Implications of Ramp Volume Estimates 
The derived ramp volume equation (restated in Figure 70) can be used to plot 
ramp volume estimates at each of the four locations in Table 22 as a function of ramp 
angle of incline (Figure 71).  We can compare reasonable ramp volumes in this graph to 
ahu volumes by considering the areas and approximate thicknesses of ahu at each 
location (Table 22, Figure 72).  In one line of reasoning, consider the approximate 
volume of the ahu at Te Pito Kura in Table 22 (1,530 m3).  If all of the volume in the 
ahu had gone into a ramp, the ramp’s angle of incline would have been approximately 
12 degrees (Figure 71).  From Figure 67, we can conclude that that this angle of incline 
requires approximately 15 people to successfully parbuckle the pukao. 
 
Figure 70.  Equation for the volume of a ramp (equation 28).  Here, volume is a 
function of the width of the ramp path (w), the combined height of the moai and ahu (h, 
refer back to Table 21), the angle of incline of the ramp ( ), and the angle of repose of 
the sloping sides of the ramp ( ).  I assume that w in each of the four cases in Table 21 
is the height of the pukao plus a margin of 60 centimeters on either side.  Regarding the 
angle of repose (considered another constant), I follow Mulloy’s assumption that 60 
degrees is reasonable (1970, 16). 
 
 
 
 
 
 
122  
ahu  Area (m2) Thickness (m) Volume (m3) Vaihu 1,440 1.5 2,160 Ura Uranga te Mahina 900 1.5 1,350 Akahanga 1,360 1.5 2,040 Te Pito Kura 1,020 1.5 1,530 
Table 22.  Estimated absolute scale values for ahu dimensions.  Dimensions from Te 
Pito Kura are taken from Heyerdahl and Ferdon, 1961, and the rest are taken from 
orthophotos developed from aerial photographs. 
 
Figure 71.  Ramp volume as a function of ramp angle of incline.  Due to the relatively 
tall moai at Te Pito Kura, the ramp at this location (1) has the largest volume for any 
given angle of incline.  Ramps at Vaihu, Ura Uranga te Mahina, and Akahanga (2,3,4) 
have similar volumes due to similar moai heights, with slight differences stemming 
only from the variable ramp path width. 
 
 
 
123  
 
Figure 72.  Scale sketch showing the extent of the ahu at Te Pito Kura (after Heyerdahl 
and Ferdon, 1961: 196).  Note the fallen moai (a) near center and the roughly 11.5 
metric ton pukao (c) off the front of the fallen moai. 
 
 
124  
Less massive pukao require smaller numbers of people for transport, and ramp 
volumes easily fit within ahu volume estimates in Table 22.  Consider Vaihu as an 
example.  With 10 people successfully pulling, an angle of incline of about 13 degrees 
or less is required (Figure 67).  The volume of such a ramp would be approximately 730 
m3 (Figure 71), which is well below the approximate ahu volume of 2,160 m3 (Table 
22).  Overall, the volume of the aforementioned ramp would spread across the area of 
the ahu at Vaihu with a thickness of about half of a meter.  For analogy, this 
corresponds to the volume of material spread with a thickness of about 14 centimeters 
across an American football field (area of about 5,351 m2).  Overall, ramp volume 
estimates support Thomson’s (1891, 449) idea regarding pukao ramps that “the stones 
which formed the foundation of the roadway were disposed of in building the wing-
extensions of the platform.” 
A final line of reasoning possible with the graphs produced in this section 
considers the remains of a possible ramp feature observed in the field at Ura Uranga te 
Mahina (Figure 73).  Specifically, we can verify that a ramp with the observed length 
would be feasible for pukao transport.  A length of about 13 meters and a moai/ahu 
height of about 8 meters dictates a ramp angle of incline of about 32 degrees.  A ramp 
with these dimensions would have had a reasonable volume (less than 500 m3, Figure 
71), and about 8,000 N of force would have been required to roll a pukao up the ramp 
through parbuckling.  This roughly corresponds to the pulling force of about a dozen 
individuals, so we find that it is physically feasible for the stone body in Figure 73 to 
represent the remnants of ramp with roughly original orientation and extent. 
 
 
125  
 
Figure 73.  Spread of stones at the ahu at Ura Uranga te Mahina measuring about 13 
meters in length (photo by Carl Lipo, 2014).  Note that pukao 29 (on the lower end of 
the possible ramp remains) has an almost perfectly circular cross section, with a 
min/max diameter of about 0.99. 
Modifications to Ramp Hypothesis 
Physically rolling a pukao up a ramp holds two advantages over sliding.  First, a 
mechanical advantage associated with parbuckling halves the force required to move a 
pukao up an incline.  Given the added force of friction associated with sliding, 
parbuckling more than halves the force required to move a pukao up an incline.  
Second, rolling takes advantage of the force of friction between the pukao and ground 
instead of fighting this force. 
 
 
126  
When a ramp is built in front of a moai, the moai may be oriented in a couple of 
different ways:  as it was when it walked (leaning forward) or as it is in its final state 
(erect).  We know that the Rapanui modified the bases of moai at some point after they 
reached the ahu in order for the moai to stand erect (Figure 74).  The forward-leaning 
orientation (Figure 75) best explains the evidence for three reasons.  First, the forward 
lean serves to slightly reduce the required height and length of the ramp.  Second, it 
serves to prevent the moai from toppling backwards due to the weight of the ramp.  
Finally, it aids the placement of the pukao on top of the moai once the former reaches 
the top of the ramp. 
  
Figure 74.  Image illustrating the convergence of two flat surfaces (marked with black 
lines) on a fallen moai base just east of Tongariki (photo by author, 2014).  In this case, 
the moai faces upwards.  The upper flat base surface in the photo may represent the 
surface when the moai leaned forward, while the lower flat base surface in the photo 
may represent the surface when the moai was in the process of final erection.  Note the 
backward/forward lean angle of approximately 23 degrees. 
 
 
 
127  
 
Figure 75.  Scale sketch showing an approximation of a reasonable ramp coming off of 
the front of a forward-leaning moai at the Vaihu ahu.  About 10 people would be 
required to pull the pukao preform up the 15 degree incline.  The top view shows the 
spread of the ramp when we consider an angle of repose about 30 degrees. 
The third point mentioned above is significant and ultimately links back to the 
presence of base indentations on pukao.  Mulloy (1970, 16) notes “the delicate problem 
of tipping the topknot” once it reaches the top of the moai.  However, the forward lean 
of the moai reduces the angle of passage for tipping.  The presence of a top cylinder 
may aid the tipping of pukao when one considers the possibility of wrapping rope 
around the top cylinder and pulling from behind the moai.  Recall that only pukao 65 
does not possess a top cylinder, but also recall that this pukao is quite conical in form.  
 
 
128  
The conical form of this pukao would serve a similar function as the presence of a top 
cylinder when tipping the pukao.  Also note that tipping over the front of the moai 
would be more stable than tipping over the side given that the former situation involves 
tipping over the short axis of the pukao oblong cross section. 
When the pukao rests on the top of the moai following tipping, we come to 
recognize a likely functional aspect of the base indentation not considered in other 
hypotheses.  Specifically, based on the forward lean of the moai, the vertical projection 
of the center of mass of the pukao moves toward the front of the pukao base indentation 
(Figure 76).  If the projection of the center of mass extends beyond the front of the 
pukao, then the pukao requires additional support to maintain its position atop the moai.  
Generally, road moai that the Rapanui walked include forward lean angles of more than 
15 degrees (Lipo et al. 2013, e.g. Figure 74).  These forward lean angles put the 
projection of the pukao center of mass close to the front edge of the base indentation 
(Figure 76 and Table 23).  In this precarious situation, one may expect external support 
to be required, as envisioned in Mulloy’s hypothesis (Figure 12).  However, the base 
indentation, and specifically its pronounced back edge (Figure 77), serve to prevent the 
pukao from toppling forward when the ramp is removed and the moai is brought to an 
erect position.  Thus, the base indentation has a function only during the final stage of 
erection.  Even though González believes that the base indentation served a function in 
balancing the pukao once it rests on top of the erect moai (Ruiz-Tagle 2005, 58), we 
know from measurements in Table 6 that this is not the case.  Overall, the presence and 
orientation of the pukao base indentations singularly support the Modified Ramp 
Hypothesis. 
 
 
129  
 
Figure 76.  Sketch of side view of moai and pukao showing the vertical projection of 
the pukao center of mass (CM) when given a forward moai lean angle of phi.  Here, phi 
is defined as the angle that places the projection of the CM directly above the front edge 
of the base indentation.  In cases where the forward moai lean angle approaches phi, the 
back edge of the base indentation comes to serve a function as a “lip” that prevents the 
pukao from toppling forward. 
 
Location Pukao Number d (m) h (m) Φ 
Hekii 57 0.177434 0.570346 17.28 
Hekii 60 0.275483 0.807894 18.83 
Anakena 65 0.239446 0.460120 27.49 
Anakena 68 0.212213 0.514347 22.42 
Table 23.  Uncalibrated measurements between pukao features and centers of mass.  
Distance d gives the horizontal distance between the center of mass and the forward 
edge of the base indentation, and distance h gives the vertical distances between the 
base of the pukao and the center of mass (refer back to Figure 35 for clarification).  
Angle phi corresponds to that in Figure 76 and is given by the inverse tangent of d/h.  
Unfortunately, the measurements in this table can only be observed in the four pukao 
above due to the poor condition, partial burial, and/or resting faces of other pukao. 
 
 
 
130  
 
Figure 77.  Textured model of pukao 57 from Hekii showing the pronounced back edge 
of its base indentation (about 4 centimeters deep) in the upper right and the relatively 
indistinct front edge of its base indentation near the center.  As seen in Table 23, pukao 
57 has a relatively low value for phi, which may explain why the back edge of its base 
indentation is relatively pronounced. 
The evidence indicates that the hypothesis for pukao transport by rolling up a 
ramp most parsimoniously explains the archaeological record.  It is also physically 
feasible.  Building a ramp would have constituted the greatest task in pukao placement.  
However, it is significant that small numbers of people likely placed pukao, consistent 
with the relatively small number of individuals required for moai transport (Lipo et al. 
2013). 
Future Research 
Future research should look more closely at spatial variability in the physical 
form of pukao.  In this study, I used sector divisions in order to make spatial 
comparisons, but any imposed divisions such as mine lead to the modifiable aerial unit 
problem.  Attempts at spatial interpolation for visualizing dimensions of pukao 
variability across space met with limited success, but ordination procedures may shed 
 
 
131  
additional light on the spatial distribution of pukao variability.  Unfortunately, the 
limited sample size of pukao in many cases may limit the detection of significant spatial 
trends in pukao variability. 
Additionally, this study and that on moai walking (Lipo et al. 2013) suggest that 
rope played an important role in transporting colossal stones for monument building on 
Rapa Nui.  However, the tensile strength of rope made of Triumfetta semitriloba is 
poorly documented, and we lack a clear understanding of the amount of plant material 
required to make a given length of rope.  Research designed to fill these gaps in our 
knowledge may shed additional light on the possible importance of Triumfetta 
semitriloba for rope production on Rapa Nui. 
Closer study of ahu may further test the previous conclusions regarding pukao 
transport that are based primarily in the study of pukao variability.  Examining the 
distribution of soil and stone cobbles in and around ahu may or may not reveal evidence 
of past ramp features.  A large complication in such a study is that stones from many 
ahu have been incorporated into walls and other features associated with past sheep 
ranching. 
Finally, work with cosmogenic dating may give us some sense of the timing of 
pukao production.  One option is the application of optically stimulated luminescence 
(OSL) dating to sediments beneath pukao and other scoria bodies.  OSL dating is based 
on the fact that ionizing radiation from isotopes of uranium, thorium, rubidium, and 
potassium is absorbed and stored in the crystal lattice of buried sediments.  When the 
crystal lattice is exposed to solar radiation, it expels its stored radiation.  The rate of 
buildup of ionizing radiation is known in certain minerals, and the buildup in a given 
 
 
132  
sample can be measured by release in controlled laboratory settings.  This enables the 
successful dating of the last time that a sample experienced sunlight.  By measuring 
OSL of sediments found beneath blocks of scoria inland of ahu and in Puna Pau, we can 
learn when these objects came to rest in their present position – the event at which 
grains on the surface became buried.  We can also use this technique to date the time 
when fallen pukao came to rest.  The former application is more likely to provide 
estimates on when the Rapanui were producing pukao.  OSL dating works well with 
quartz and feldspars.  Given the mafic nature of the Rapanui substrate, OSL dating must 
in this case rely on feldspars (primarily plagioclase) present in buried sediments. 
Another form of cosmogenic dating that can be applied to pukao movement is 
cosmic ray exposure (CRE) dating.  This technique takes advantage of the known rates 
of decay for certain cosmogenic radionuclides such as beryllium-10 and helium-3.  An 
examination of helium-3 may be the most fruitful on Rapa Nui given that this isotope is 
frequently reported in olivine and pyroxene (both mafic minerals, Gosse and Phillips 
2001, 1516).  Unfortunately, most successful applications of helium-3 CRE dating have 
been on the scale of thousands of years (Figure 78, Gosse and Phillips 2001, 1517). 
 
 
133  
 
Figure 78.  Various dating methods and their effective ranges (after Pánek, 2015: 170).  
Dating methods with clear ties to puko production and effective ranges between 0.1 to 1 
ka are most likely to expand our understanding of a chronology for pukao production.  
Lichenometry has a relatively recent effective range when compared to the OSL and 
CRE dating methods. 
 
A final possible method for dating pukao production and transport involves 
lichenometry.  Lichenometry takes various approaches, but it generally dates the 
exposure of a rock surface by considering the nature and extent of lichen growth and 
estimating past lichen growth rates based on present growth rates.  One limitation in 
applying lichenometry to rock surfaces on Rapa Nui is that very few dated reference 
lichen substrates exist.  However, Gerhard Follman (1961) successfully used 
lichenometry in order to date moai and ahu surfaces, and Rutherford et al. (2008) have 
since provided some documentation of lichen growth at Ahu Tahira.  Lichenometry has 
 
 
134  
been used to date the initial creation of ahu surfaces with ages greater than 400 years 
and errors less than 10 percent (Rutherford et al. 2008, 43).  No study has yet examined 
lichens on pukao, but Figures 25 and 27 establish that lichen has colonized pukao 
surfaces for at least one hundred years. 
Summary 
The Modified Ramp Hypothesis for pukao transport best explains the 
archaeological record and is physically feasible.  This form of transport has never 
before been proposed, but it includes elements of the traditional Ramp Hypothesis and 
Mulloy’s Simultaneous Set Up Hypothesis.  The Modified Ramp Hypothesis first 
involves the construction of a ramp (ranging in angle of incline between 10 and 35 
degrees) off the front of a forward-leaning moai after the moai is brought to stand on an 
ahu.  Ramp material, which can be used to help walk the moai up a small height to the 
top of the ahu, can include stone, coral fragments, and soil.  The pukao preform is rolled 
up the ramp primarily by a group of up to 15 individuals parbuckling from above with 
possible levering aid by up to three people from below.  Breaks or steps in the ramp 
may ease this stage of transport.  Upon reaching the top of the ramp, the preform is 
pivoted in place by excavating the ramp material beneath the preform, manipulating the 
preform through levering, and/or manipulated the preform by using rope.  The forward 
lean of the moai eases the tipping of the preform so that the base indentation fits on the 
moai’s top.  Any refinishing of the pukao occurs at this point.  Material is then carved 
from the rear of the moai’s base so that the moai tilts backward and is brought to its 
final erect position.  Ropes may stabilize the moai during this transit, and the pukao 
base indentation prevents the pukao from toppling forward.  The final step is the 
 
 
135  
removal of the ramp: Stones can be incorporated into the wings of the ahu, and soil can 
be used in nearby garden plots. 
Future research may refine aspects of the above reconstruction for pukao 
transport.  However, it is significant that the archaeological evidence for pukao 
transport, just like that for moai transport (Hunt and Lipo 2011), supports a method 
involving relatively few people.  Contrary to popular ideas that connect monumentality 
with larger groups of laborers working under a centralized authority, neither moai nor 
pukao transport required the coordination of large groups of people. 
 
  
  
Appendix 
Year Name(s) Purpose of Visit 
1722 Jacob Roggeveen 
Cornelis Bouman 
Dutch expedition in search 
of Terra Australis 
1770 F.A. de Agüera y Infanzon 
Felipe González de Haedo 
Spanish territorial mission 
1774 James Cook 
Johann and Georg Forster 
British exploratory mission 
1786 Jean François Galaup de la 
Pérouse 
French exploratory mission 
1866 Hippolyte Roussel French missionary 
1868 Linton Palmer British survey expedition 
1877 Alphonse Pinart French exploratory 
expedition 
1882 Wilhelm Geiseler German anthropological 
expedition 
1886 William Thomson Smithsonian expedition 
1914 Katherine and Scoresby 
Routledge 
British private / museum-
based expedition 
1940 Alfred Métraux French ethnographic 
expedition 
1955 Thor Heyerdahl Norwegian archaeological 
expedition 
Appendix Table 1.  List of past visiting groups with references to pukao.  Jacob 
Roggeveen is credited with the European discovery of Rapa Nui. 
Model Generation, Filing, and Observation Strategies 
Refined workflow for pukao textured model generation: 
• Align photos with high accuracy and disabled preselection (typically takes between 3 
and 13 hours) 
• Optimize alignment (typically takes 5 minutes) 
• Compute alignment quality for individual photos 
• Disable cameras that have quality values lower than 0.5 
• In situations with a multitude of photos, disable duplicate cameras and thin those with 
quality values lower than 0.7 
• Save sparse point cloud 
• Generate dense point cloud with high quality and aggressive filtering (Note that 
aggressive filtering eliminates some detail by filtering outliers, but with the scale of the 
imagery in use, aggressive is okay - typically takes between 3 and 7 hours.) 
• Save dense point cloud 
• Generate mesh based off of arbitrary surface type and dense point cloud (typically takes 
around 1 hour) 
• Build model texture using generic approach (typically takes less than an hour) 
 
 
137  
• Save entire project 
• Generate and save accuracy report 
• Backup all files on computer and external hard drive 
 
Model file management scheme: 
“Pukao 2014 Project” 
>Place Name (e.g. “Vinapu”) 
  >“Photos” 
>Pnumber (e.g. “P2,” all photos of individual numbered pukao used in 
model)… 
   >“Generic” (overview photos of site) 
  >Pnumber (e.g. “P2,” file including all model files of an individual pukao) 
   >“Unclipped” 
    >Photoscan project file (e.g. “Pukao2.psz”) 
    >Project report (e.g. “Pukao2.pdf”) 
    >Exported point file (e.g. “Pukao2” .obj file) 
    >Exported model file (e.g. “Pukao2mod” .obj and .mtl files) 
   >“Clipped” 
>Photoscan project file (e.g. “Pukao2clip.psz”) 
    >Project report (e.g. “Pukao2clip.pdf”) 
    >Exported point file (e.g. “Pukao2clip” .obj file) 
>Exported model file (e.g. “Pukao2modclip” .obj and .mtl 
files) 
 
Preliminary Analysis of Textured Models: 
In Photoscan: 
• Record the resting position of the pukao. 
• Record whether or not the full pukao is exposed. 
• Record the state of the pukao. 
• Clip model so that it includes only points associated with the pukao. 
• Fill 100% of the holes in the pukao model. 
• Record the volume of the pukao and associated error 
In Meshlab: 
• Apply the APSS colorize curvature filter to the mesh using the default settings – This 
colorizes the vertices of the mesh using the curvature of the underlying surface and 
enhances subtle and narrow etchings in the pukao surface. 
• Use the z-painting tool in order to mark etchings and indentations in the pukao surface 
in red.  Only markings that appear to have been created outside of weathering should be 
marked.  Export the marked mesh as a .obj file.  (In the filing scheme previously given, 
the marked .obj files go into the “clipped” file associated with the given pukao.) 
• Using the PickPoints tool, select each marking and assign it a unique number going 
upward from 1 on each pukao.  Save this information as a .pp file with the 
aforementioned .obj file.   
• After marking and labeling all features, record the presence and absence features and 
take measurements of key dimensions. 
 
 
138  
Additional Physics Derivation: Final Flipping of Pukao 
We wish to examine the force required to flip the pukao over once it reaches the top of 
the ramp and is aside the top of the moai as shown below in Appendix Figure 1. 
 
 
Appendix Figure 1.  A possible means of flipping a pukao from its side onto its base 
using a rope tied around the top cylinder and trailing off the side of the moai.  The force 
on the pukao due to gravity acts downward at the center of mass of the pukao and is 
given by mg, where m is the mass of the pukao and g is the acceleration due to gravity.  
The normal force due to the ramp acts perpendicular to the ramp and is designated by 
.  A person exerts a force  that acts at an angle  above the top surface of the 
moai (of length b).  In this case,   pivots the pukao around the pivot point S.  The 
lever arm of  is p, and the lever arm of mg is c, as seen above.  Note that the pukao 
may be flipped over the front or the side of the moai depending on where the ramp is 
located.   
 
We wish to find an expression that gives the force  required for pivoting the pukao 
round the point S.   We begin by considering the situation in which the pukao is in a 
state of static equilibrium and the sum of torques is zero.  We thus have equation (29), 
in which  is the component of  acting along the x-direction and perpendicular to 
the lever arm of length p. 
 
  (29) 
 
 
 
139  
We know that  is related to  through equation (30). 
 
  (30) 
 
Thus, we can substitute equation (30) into equation (29), solve for , and come to 
equation (31). 
 
 
 
(31) 
   
Note that we can find equation (32), which relates known lengths to .  
 
 
 
 
 (32) 
  
 
By substituting equation (32) into equation (31), we come to our final result given by 
equation (33). 
 
 
 
 
 (33) 
  
 
Using equation (33) and applying a given pukao mass (m), distance to pukao center of 
mass (c), width of pukao being flipped (p), and base of moai top (b), we can now find 
the force  required to flip the pukao.   
 
 
140  
Bibliography 
Agüera y Infanzon, F. A. de. Journal of the principal occurrences during the Voyage of 
the Frigat ‘Santa Rosalia’ in the year 1770. Hakluyt Society, 2nd ser., no. 13. 
Cambridge, 1908.  (original work produced in 1770) 
Allen, Melinda S. "Style and function in East Polynesian fish-hooks." Antiquity70.267 
(1996): 97-116. 
Altman, A.M., and F. Morin. Early Visitors to Easter Island, 1864-1877: The Reports of 
Eugene Eyraud, Hippolyte Roussel, Pierre Loti, and Alphonse Pinart. Easter 
Island Foundation, Bearsville Press, 2004. 
Ayres, William S., and G. S. Ayres. "Geiseler's Easter Island Report: An 1880s 
Anthropological Account." Asian and Pacific Archaeology Series 12 (1995). 
Bahn, Paul. “The Archaeology of the Monolithic Sculptures of Rapanui: A General 
Review.” Easter Island Studies: Contributions to the History of Rapanui in 
Memory of William T. Mulloy. Ed. Steven R. Fischer. Oxbow Books Limited, 
1993. 82-85. 
Baker, Peter E. "Preliminary account of recent geological investigations on Easter 
Island." Geological Magazine 104.02 (1967): 116-122. 
Bandy, Mark C. "Geology and petrology of Easter Island." Geological Society of 
America Bulletin 48.11 (1937): 1589-1610. 
Balfour, Henry. "The statues of Easter Island." Folklore (1921): 70-72. 
Beaumont, Hubertus JE, Jenna Gallie, Christian Kost, Gayle C. Ferguson, and Paul B. 
Rainey. "Experimental evolution of bet hedging." Nature 462.7269 (2009): 90-
93. 
Brown, John Macmillan. The riddle of the Pacific. Adventures Unlimited Press, 1996. 
Charola, A. Elena. Easter Island: the heritage and its conservation. World Monuments 
Fund, 1994. 
Charola, A. Elena. Death of a moai: Easter Island statues: their nature, deterioration 
and conservation. Bearsville Press and Cloud Mountain Press; Easter Island 
Foundation, 1997. 
Cook, James. A Voyage towards the South Pole, and round the World, performed in 
H.M.SS. “Resolution” and “Adventure,” in the years 1772-75. 2 vols. London, 
1777. 
Diamond, Jared. Collapse: How societies choose to fail or succeed. Penguin, 2005. 
 
 
141  
Dunnell, Robert C. "Seriation method and its evaluation." American Antiquity (1970): 
305-319. 
Dunnell, Robert C. "Style and function: a fundamental dichotomy." American 
Antiquity (1978): 192-202. 
Dunnell, Robert C. "Methodological issues in Americanist artifact 
classification." Advances in Archaeological Method and Theory (1986): 149-
207. 
Edwards, Edmundo, Raul Marchetti, Leopoldo Dominichetti, and Oscar Gonzáles-
Ferrán. “When the Earth Trembled, the Statues Fell.” Rapa Nui Journal 10.1 
(1996): 15-21. 
Fischer, Steven R. Island at the End of the World. Reaktion Books Ltd., 2005. 
Follmann, Gerhard. “Estudios liquenometrics en los monumentos prehistoricos de la 
Isla de Pascua.” Rev. Universitaria (Univ. Catolica de Chile) 46 (1961): 49-154. 
Gosse, John C., and Fred M. Phillips. "Terrestrial in situ cosmogenic nuclides: theory 
and application." Quaternary Science Reviews 20.14 (2001): 1475-1560. 
Gurley, Robert E., and William Liller. “Palm Trees, Mana, and the Moving of the 
Moai.” Rapa Nui Journal 11.2 (1997): 82-84. 
Hamilton, Sue. "Rapa Nui Landscapes of Construction." Archaeology International 10 
(2007). 
Hamilton, Sue. "Rapa Nui (Easter Island)’s Stone Worlds." Archaeology 
International 16 (2013): 96-109. 
Hamilton, Sue, Mike Seager Thomas, and Ruth Whitehouse. "Say it with stone: 
constructing with stones on Easter Island." World Archaeology 43.2 (2011): 
167-190. 
Herrera, Christian, and Emilio Custodio. "Conceptual hydrogeological model of 
volcanic Easter Island (Chile) after chemical and isotopic surveys." 
Hydrogeology Journal 16.7 (2008): 1329-1348. 
Heyerdahl, Thor, and Edwin N. Ferdon. Archaeology of Easter Island. Stockholm: 
Forum Pub. House, 1961. 
Heyerdahl, Thor. Easter Island: the mystery solved. New York: Random House, 1989. 
Horley, Paul. “Identification of Sites Illustrated in the Easter Island Report by William 
J. Thomson.” Rapa Nui Journal 23.1 (2009): 12-17. 
 
 
142  
Horrocks, M., et al. "Microfossils of Polynesian cultigens in lake sediment cores from 
Rano Kau, Easter Island." Journal of Paleolimnology 47.2 (2012): 185-204. 
Hunt, Terry L., and Carl P. Lipo. "Late colonization of Easter Island." Science 311.5767 
(2006): 1603-1606. 
Hunt, Terry, and Carl Lipo. The statues that walked: unraveling the mystery of Easter 
Island. Simon and Schuster, 2011. 
Im Thurn, Everard. “The island of the stone statues.” Nature 105 (1920): 583-584. 
Kahn, R. L. "Georg Forsters Werke Samtliche Schriften, Tagebucher, Briefe. A Voyage 
Round the World." (1968). 
Kersten, Thomas P., Maren Lindstaedt, and Burkhard Vogt. "Preserve the Past for the 
Future–Terrestrial Laser Scanning for the Documentation and Deformation 
Analysis of Easter Island's Moai." Photogrammetrie-Fernerkundung-
Geoinformation 2009.1 (2009): 79-90. 
Kroemer, KH Eberhard. "Human strength: terminology, measurement, and 
interpretation of data." Human Factors: The Journal of the Human Factors and 
Ergonomics Society 12.3 (1970): 297-313. 
Lee, Georgia. Rock art of Easter Island: symbols of power, prayers to the gods. Vol. 17. 
Institute of Archaeology University of California Lo S, 1992. 
Lee, Georgia. “Petroglyph Motif Distribution in East Polynesia.” Rapa Nui Journal 
11.1 (1997): 5-9.   
Lee, Vincent R. “Awakening the Giant.” Rapa Nui Journal 26.2 (2012): 5-15. 
Lipo, Carl P., and Mark Madsen. "Neutrality,‘‘style’’, and drift: building methods for 
studying cultural transmission in the archaeological record." Style and Function: 
Conceptual Issues in Evolutionary Archaeology. Bergin and Garvey, Westport, 
Connecticut (2001): 91-118. 
Lipo, Carl P., Terry L. Hunt, and Brooke Hundtoft. "Stylistic variability of stemmed 
obsidian tools (mata’a), frequency seriation, and the scale of social interaction 
on Rapa Nui (Easter Island)." Journal of Archaeological Science 37.10 (2010): 
2551-2561. 
Lipo, Carl P., and Terry L. Hunt. "Mapping prehistoric statue roads on Easter Island." 
Antiquity 79.303 (2005): 158-168. 
Lipo, Carl P., Terry L. Hunt, and Sergio Rapu Haoa. "The ‘walking’ megalithic statues 
(moai) of Easter Island." Journal of Archaeological Science 40.6 (2013): 2859-
2866. 
 
 
143  
Lipo, Carl P., et al. "Population structure, cultural transmission, and frequency 
seriation." Journal of Anthropological Archaeology 16.4 (1997): 301-333. 
Love, Charles M. "More on moving Easter Island Statues, with Comments on the 
NOVA Program." Rapa Nui Journal 14.4 (2000): 115-118. 
Martinsson-Wallin, Helene, and Susan J. Crockford. "Early settlement of Rapa Nui 
(Easter Island)." Asian Perspectives 40.2 (2001): 244-278. 
Métraux, Alfred, and Michael Bullock. Easter Island: a stone-age civilization of the 
Pacific. Book Club Associates, 1957. 
Métraux, Alfred. Ethnology of Easter Island. No. 160. Bishop Museum Pr, 1971. 
(original work published 1940) 
Mudg, M., et al. "Principles and practices of robust, photography-based digital imaging 
techniques for museums." (2010). 
Mulloy, William. "A Speculative Reconstruction of Techniques of Carving 
Transporting and Erecting Easter Island Statues." Archaeology & Physical 
Anthropology in Oceania (1970): 1-23. 
O’Brien, Michael J., and Robert D. Leonard. "Style and function: an introduction." Style 
and Function: Conceptual Issues in Evolutionary Archaeology (2001): 1-23. 
Orliac, Catherine, and Michel Orliac. Easter Island: Mystery of the Stone Giants. HN 
Abrams, 1995. 
Palmer, John Linton. "Observations on the inhabitants and the antiquities of Easter 
Island." Journal of the Ethnological Society of London (1869-1870) (1869): 371-
377. 
Palmer, J. Linton. "A visit to Easter Island, or Rapa Nui, in 1868." Journal of the Royal 
Geographical Society of London (1870): 167-181. 
Pánek, Tomáš. “Recent progress in landslide dating: A global overview.” Progress in 
Physical Geography 39.2 (2015): 168-198. 
Passos, John Dos. Easter Island: Island of Enigmas. Garden City, NY: Doubleday, 
1971. 
Pavel, Pavel. “Reconstruction of the transport of the moai statues and pukao hats.” Rapa 
Nui Journal 9.3 (1995): 69-72. 
Richards, Rhys. Easter Island 1793 to 1861: Observations by Early Visitors before the 
Slave Raids. Easter Island Foundation, 2008. 
 
 
144  
Rivers, W. H. R. "The statues of Easter Island." Folklore 31.4 (1920): 294-306. 
Routledge, Katherine. The Mystery of Easter Island. Adventures Unlimited, 1998. 
(original work published 1919) 
Routledge, Scoresby. "Easter Island." Geographical Journal (1917): 321-340. 
Ruiz-Tagle, Eduardo. "Easter Island: the first three expeditions." Museum Store, 
Rapanui Press: Rapa Nui (2005). 
Rutherford, Susannah, Brett Shepardson, and Jesse Stephen. “A Preliminary 
Lichenometry Study on Rapa Nui – The Rapa Nui Youth Involvement 
Program.” Rapa Nui Journal 22.1 (2008): 40-47. 
Saher, Herbert von. "The complete journal of Captain Cornelis Bouman, Master of the 
ship Theinhoven, forming part of the fleet of Jacob Roggeveen, from 31 March 
to 13 April 1722 during their stay around Easter Island." Rapa Nui Journal 8.4 
(1994): 95-100. 
Seager Thomas, Mike. "Stone use and avoidance on Easter Island: Red scoria from the 
topknot quarry at Puna Pau and other sources." Archaeology in Oceania (2014). 
Skinner, H. D. "The Easter Island Figures." Folklore 33.3 (1922): 296-299. 
Snook, Stover H., and Vincent M. Ciriello. "The design of manual handling tasks: 
revised tables of maximum acceptable weights and forces." Ergonomics 34.9 
(1991): 1197-1213. 
Steadman, David W., Patricia Vargas Casanova, and Claudio Cristino Ferrando. 
"Stratigraphy, chronology, and cultural context of an early faunal assemblage 
from Easter Island." Asian Perspectives 33.1 (1994): 79-96. 
Stocks, Denys A. Experiments in Egyptian archaeology: stoneworking technology in 
ancient Egypt. Routledge, 2013. 
Thomson, William Judah. Te Pito Te Henua; Or, Easter Island. Smithsonian Institution, 
1891. 
Treister, Kenneth, Patricia Vargas Casanova, and Claudio Cristino. Easter Island's 
Silent Sentinels: The Sculpture and Architecture of Rapa Nui. UNM Press, 2013. 
Van Tilburg, J. 1986. Red scoria on Easter Island. Journal of New World Archaeology, 
7(1): 1–27. 
Van Tilburg, Jo Anne. "HMS Topaze on Easter Island." British Museum, London 
(1992). 
 
 
145  
Van Tilburg, Jo Anne, and Georgia Lee. "Symbolic stratigraphy: Rock art and the 
monolithic statues of Easter Island." World Archaeology 19.2 (1987): 133-149. 
Von Däniken, Erich. Chariots of the Gods?: Unsolved Mysteries of the Past. New York: 
Putnam, 1970. 
Wilmshurst, J.M., T.L. Hunt, C.P. Lipo, and A.J. Anderson. "High-precision 
radiocarbon dating shows recent and rapid initial human colonization of East 
Polynesia." Proceedings of the National Academy of Sciences 108.5 (2011): 
1815-1820. 
Wozniak, Joan A. "Prehistoric horticultural practices on Easter Island: lithic mulched 
gardens and field systems." Rapa Nui Journal 13.3 (1999): 95-99.