Journal of Pacific Archaeology – Vol. · No. ·
– –
North Beretania St., Apt. , Honolulu, HI
Anthropology Department, College of William & Mary,
Washington Hall, Room , Jamestown Rd.,
Williamsburg, VA
* Corresponding author: tsd@tsdye.online
Submitted //, accepted //. First online //.
Functional Classification of Hawaiian
Curved-Edge Adzes and Gouges
omas S. Dye¹
*
& Jennifer G. Kahn²
As part of a project to describe and classify functionally more than Hawaiian stone adzes held in the ethnographic
and archaeological collections at Bishop Museum in Honolulu, tools with curved edges were identified and described.
e curved-edge tools include adzes and gouges, which can be unambiguously distinguished from one another using a
combination of weight and length index. Many of the curved-edge adzes have large cutting edge width ratios; the nar-
row shoulders and wide edges led archaeologists to describe them as ‘hoofed’. Curved-edge adzes and gouges make up
less than the percent of the Hawaiian collection. eir rarity in Hawai‘i appears to be in line with other island groups
in East Polynesia outside New Zealand, where they make up about percent of museum collections.
Keywords: Hawai‘i, adze, gouge, classification
As part of a project to describe and classify functionally
more than Hawaiian stone adzes held in the ethno-
graphic and archaeological collections at Bishop Museum
in Honolulu, this paper identifies and describes tools
with curved edges. e project has several interrelated
goals, among them: i) develop an alternative to a current
practice of Hawaiian archaeology that identifies and dis-
tinguishes adzes by their transverse sections; ii) enhance
analytic replicability by defining true classes distinguished
by their boundaries, rather than groups distinguished by
central tendencies; iii) complement the progress made
in identifying adze making habits through the study of
reduction sequences and replication experiments with a
more complete description of the variability inherent in
the finished product; and iv) develop a descriptive focus
on function rather than culture history.
When Hawaiian adzes are grouped according to the
Duff categories (Duff ), the uniformity of their quad-
rangular transverse sections stands out in comparison to
adzes from other Polynesian island groups, where oval,
lenticular, plano-convex, trapezoidal, sub-triangular, and
triangular transverse sections are commonly found (Emory
). is is a desirable quality in regional comparison, but
within Hawai‘i the importance of transverse section in the
Duff categories and the striking predominance of quadran-
gular sections work in tandem to lump the adzes together,
rather than distinguishing them in ways that might be
analytically useful.
e high level of skill needed to manufacture a stone
adze, and the typical Hawaiian quadrangular-section adze
in particular, led archaeologists to study the reduction se-
quence practiced by the adze maker from the evidence of
waste materials at quarry sites and by carrying out repli-
cation experiments (Cleghorn ). is line of inquiry
has made impressive progress. Based on a study of manu-
facturing waste from three quarries on the northwest end
of Moloka‘i Island, a detailed description of the reduction
sequence documents the order in which characteristic adze
features – bidirectional edges required for the quadrangu-
lar cross section, bevel, poll, and tang – were craed by the
adze maker (Clarkson et al. ). Replication experiments
prove that the quadrangular-section Hawaiian adze can be
made with ‘direct hard and so stone hammer percussion’
(Clarkson et al. : ) in a sequence of flaking deter-
mined by the form of the raw material, either a cobble, a
flake, or a piece of tabular stone. e habits developed by
expert Hawaiian adze makers can now be appreciated in
broad outline, although the finesse with which they of-
ten worked has so far escaped replication (Clarkson et al.
: ).
In contrast, the last attempt at a comprehensive de-
scription of Hawaiian adzes was completed more than a
century ago (Brigham ). e canonical description of
Hawaiian material culture does not include adzes, presum-
ably because the author died before work with the adzes
was complete (Buck ). Other work with the adze collec
-
tions at Bishop Museum focuses on the transverse section
Journal of Pacific Archaeology – Vol. · No. ·
and presents results statistically, rather than descriptively
(Duff ; Emory ).
e design of the project reported here is based on the
results of carving experiments with Māori adzes in New
Zealand that identified functional differences among the
Duff categories commonly used by Pacific archaeologists
(Turner , ; Turner and Bonica ). Because
culture history requires artifact classes based on stylis-
tic, rather than functional, attributes (Dunnell ), the
limits of the culture historical project envisioned by Duff
and by subsequent archaeologists who have described and
grouped Polynesian adzes with reference to the Duff cat-
egories were exposed. Nevertheless, recognition of these
limits to the culture history project should not obscure
the fact that the Duff categories do tend to group like with
like and that archaeologists continue to ‘recognise some
practical strengths in the system’ (Shipton et al. : ).
Archaeologists today increasingly regard the Duff cate-
gories as groups that represent one step along a path to
classifications (Dunnell , ) that might be used to
structure inquiries into topics such as the technology of
adze manufacture (e.g., Shipton et al. ) or the history of
regional interaction spheres based on geochemical sourc-
ing programs (e.g., Richards ).
is paper is the second in a series designed to de-
scribe the formal variability of complete Hawaiian adzes in
the ethnographic and archaeological collections of Bishop
Museum and to classify them according to function. e
first paper in the series identified Hawaiian stone axes
that resemble the Māori adzes grouped by Duff as Type
(Kahn and Dye ). e axes are designed to be ‘lateral-
ly-haed excavation tools for working in confined spaces’
(Turner : ). Kahn and Dye () identify the double
bevel as the characteristic that distinguishes them from
the more common single-bevel tools and leads to their
classification as axes. ese tools are rare wherever they
are found; they account for about percent of the New
Zealand assemblage (Turner : ) and about percent
– reported incorrectly as about . percent (Kahn and Dye
: ) – in the Hawai‘i assemblage.
Identification of a curved edge as a distinguishing
characteristic of the class extends the focus on the working
edge of the tool, whose functional importance was identi-
fied and experimentally demonstrated some time ago (Best
). e larger members of the curved edge adze class
resemble Māori adzes grouped by Duff as Type , whose
distinctive characteristics have been described as exhibit-
ing ‘frontal convexity and a scooped curved blade’ (Turner
: ). e larger curved-edge adzes were ‘designed for
shaping curved surfaces such as those found on canoe hulls
and bowls’ (Turner : ). e smaller tools resemble
Māori tools grouped as Type , which includes typically
small, slender tools oen identified as chisels or gouges.
ese tools were designed to carry out ‘a wide range of
generally detailed intricate tasks’ (Turner : ), such
as making lashing holes in canoe hulls. us, classification
of Hawaiian adzes by edge shape crosscuts the distinction
oen made between adzes and gouges.
In the following sections, the Hawaiian curved-edge
adzes are described, their representation in Hawai‘i is com-
pared to similar tools found elsewhere in Polynesia, and
their attributes are compared to the Hawaiian straight-
edged adzes, whose description and classification will be
the subject of the projected third paper in this series.
-
e cutting edge of an adze can be curved in one or both
of two ways, which have been called curve A and curve B
(Blackwood ). Edges with curve A look curved when
the tool is viewed in longitudinal plan, while those with
curve B look curved when the tool is viewed from the
distal end (fig. ). Edges with curve A are relatively com-
mon among Hawaiian adzes, typically because the edge
of the tool curves back at one or both sides. An edge with
curve A might be due to tool design or to the practice of
resharpening adzes whose once straight edges were dam-
aged at one or both ends. Edges with curve B are rarer than
those with curve A, nevertheless, edges with curve B were
almost certainly created by design, rather than from use
modification. e Hawaiian adzes that exhibit a curve
B edge are referred to here as curved-edge adzes. Although
adzes with a curve B edge frequently also exhibit curve A,
the presence or absence of curve A is not a classification
criterion.
e adzes described here are identified as Hawaiian
in Bishop Museum records, which oen rely on informa-
tion from donors that cannot be verified. e tools are
identified by their museum labels, which in most cases are
written in India ink directly on the tool. In some cases, a
paper label is affixed to the tool. Some tools exhibit multi-
ple labels, oen with different identifiers; in these cases it
is usually possible to identify the most recent label, which
Figure 1. Adze cung edge curves: a, curve A viewed in
longitudinal plan; b, curve B viewed from the distal end with
the adze front superior.
Dye & Kahn – Functional Classification of Hawaiian Curved-Edge Adzes and Gouges
is the one used to identify the tool. e identifiers on the
adze labels vary in their format; those reported here have
been normalized to reflect the convention followed in the
digital catalog at Bishop Museum, but the task of joining
the table of adze data used in this analysis with the digital
catalog is not complete. Nevertheless, museum staff recog-
nize this variability and are able to retrieve adzes with the
identifiers as reported.
Five of the Hawaiian adzes are described and il-
lustrated below. e other curved-edge adzes are de-
scribed in the Supplementary Material (Kahn and Dye
). An unusual curved-edge adze, ., has been
fully described and sourced to a quarry on the south coast
of Moloka‘i Island (Sinton and Sinoto ). is the long-
est Hawaiian adze at Bishop Museum.
In the following descriptions, terms for the parts of an
adze follow Buck et al. (), with the exception that the
edge is sometimes referred to as the cutting edge for clarity.
Terms for orientation follow the conventions established
by Garanger ().
Curved-edge Adze, B.
is hoof-shaped adze with a plano-convex transverse
section has a reduced, triangular-section butt that shows
traces of light polishing, but whose surface is primarily
flake scars (fig. a). e poll is irregular and in plan does
not form a right angle with the longitudinal section. e
color of the rock is gray (N/) and the tool has a reddish
yellow (. YR /) patina common on artifacts found on
the surface where there is red soil. e blade is fully pol-
ished on the front and the back. e edge is curved in both
A and B sections, consistent with its classification as hoof-
shaped. It is .cm long, . × .cm at mid-section, and
weighs g. e cutting edge measures .cm.
Curved-edge Adze,
is untanged, plano-convex section, very dark gray adze
appears to have been re-worked extensively to narrow the
butt (fig. b). e front and sides of the blade and the bevel
are polished, but the rest of the back and the butt are cov-
ered with flake scars with small remnant patches of polish
Figure 2. Curved-edge adzes and gouge: a, curved-edge adze, B.01799; b, curved-edge adze, 11018; c, curved-edge adze,
1970.010.020; d, curved-edge adze, C.08290; e, gouge, 70.286.21.
Journal of Pacific Archaeology – Vol. · No. ·
in between scars. e poll is flat but not polished. e tool
is widest at the edge and narrows markedly toward the poll.
It is .cm long, . × .cm at mid-section, and weighs
g. e cutting edge measures .cm.
Curved-edge Adze, ..
is tanged, triangular section, dark gray adze is polished
on the front, back, and sides of the blade (fig. c). e sur-
face of the butt has not been polished. e poll is squared
off. Shallow flakes have been removed from the front, sides,
and back of the blade, probably through use. Four small
flake scars originate at the edge. e edge is the widest part
of the tool. It is .cm long, . × .cm at mid-section,
and weighs g. e cutting edge measures .cm.
Curved-edge Adze, C.
is untanged, plano-convex section, dark grayish brown
adze is polished on most of the front, on the sides near the
edge, and on the bevel (fig. d). Flakes have been taken off
both sides to narrow the butt. e front is reduced near the
poll, which is small and irregular. Both the front and back
of this tool are curved. It is .cm long, . × .cm at
mid-section, and weighs g. e cutting edge measures
.cm.
Gouge, ..
is untanged, irregularly hexagonal section, dark grayish
brown adze is polished all over, including the poll (fig. e).
It has the slender shape of a gouge. ere are two small
flake scars on one side and one on the poll. e bevel is
convex. ere is a dark deposit from an old museum label
on the front near the poll. It is .cm long, . × .cm at
mid-section, and weighs g. e cutting edge measures
.cm.
e project goal to devise a classification of Hawaiian adz-
es that illuminates their function in old Hawai‘i requires
consideration of the concept of function, which carries a
variety of meanings and connotations. It is important to
note at the outset that a function is assigned to an adze
and is not something that the adze has independent of the
assignment. Another way to say this is that adze function
is relative to the observer (Searle : –). One implica-
tion of this is that the function assigned to an adze by the
persons who made and used it, who have long since passed
away, cannot now be known. Instead, the archaeologist
today assigns a function to an adze based on the form and
physical qualities of the tool and assumptions about the
range of materials the adze might have worked. e deci-
sion to focus on the adze edge as a primary criterion in the
classification reflects the degree to which edge character-
istics determine the effect an adze has when it is applied
with force to a piece of wood (Best ). e effect of a
curved-edge adze when it is applied to a piece of wood
can be predicted by a mechanical analysis (Cotterell and
Kamminga ), or it can be determined experimentally
(e.g., Turner ). In either case, it will be found that its
effect is to shape a curved surface, perhaps to form a groove
or cut a perforation. Having predicted and observed the
effect of a curved-edge adze when it is applied to a piece of
wood, it seems a short step to assign this effect to the func-
tion of the adze and to say that the function of a Hawaiian
curved-edge adze is to shape a curved surface in wood.
is short step assigns what Aristotle referred to as telos,
or final cause, to the production of tools assigned to the
curved adze class. Such a teleological argument appears to
imply that the function predicted by mechanical analysis
and determined experimentally was also assigned by the
adze maker and user, which would violate the claim that
adze function is relative to the observer. Fortunately, the
teleological argument does not require the final cause to
be a conscious goal (Hulswit ) that the adze maker
and user assigned as a function of adzes in the curved-
edge class. e claim that the function of a curved-edge
adze is to shape a curved surface in wood is relative to the
archaeological observer whose goal is to explain the formal
variability of Hawaiian adzes in terms of the habits of the
adze makers and users responsible for creating that vari-
ability. is is not to suggest that the habits of Hawaiian
adze makers and users were not guided by consciously
held goals. ey certainly were, but the nature of the goals,
how they related to habits, and how they fit into more ex-
tensive theories of being and acting in the world are not
illuminated by the archaeologist’s assignment of function.
e best that can be done along these lines today is to note
that there is some published evidence that Hawaiian adze
makers and users likely distinguished the larger curved-
edge adzes as ko‘ i ‘auwaha ; one English gloss for this term
is ‘scoop adze’ (Holmes : ).
Another consideration in devising a classification
of Hawaiian adzes that illuminates their function in old
Hawai‘i relates to how the tool was applied to the work.
e curved-edge adzes at Bishop Museum display a wide
variety of forms, ranging from the longest Hawaiian adze
in Bishop Museum, ., to several diminutive tools
traditionally referred to as chisels or gouges, a gouge being
a variant of the general class chisel and characterized by
a curved edge. Gouges and chisels are distinguished from
adzes by the manner in which they are applied to the work.
e artisan using a gouge or a chisel rests it against the
work and applies force either manually or by striking the
proximal end of the tool – either the blade itself or a handle
to which the blade has been attached more or less inline –
with a mallet. In contrast, an adze attached at an angle to a
handle attacks the work by percussion, where the force of
a glancing blow is supplied by the arm motion required to
move the haed tool in a swinging arc. ese differences in
Dye & Kahn – Functional Classification of Hawaiian Curved-Edge Adzes and Gouges
how the tool is attached to a handle and applied to the work
cannot be observed by the archaeologist working with an
unhaed stone tool. Nevertheless, some of the curved-
edge adzes might have been used as gouges, either with
or without a handle (Holmes : ). Although we know
of no examples of ancient Hawaiian adzes lashed in-line
to a handle, like a modern gouge, examples of small tools
lashed in-line to handles are known from elsewhere in
Polynesia, including a chisel lashed to an intricately carved
handle from the Marquesas (Linton Plate ) and
two bone chisels collected by Cook in the Society Islands
(omas et al. : , ). Archaeologists sometimes
attempt to determine whether or not a tool was used with
a ha by examining the poll to identify either ha polish
(Turner : ) or mallet abrasions (Stokes : ),
but there is no guarantee that a tool was used in just one
way. In fact, a given tool might have been applied to the
work in various ways, depending on circumstances. Using
the distinction as a basis for classification would lead to a
difficult situation where two researchers who differ in their
determinations of whether a tool might have been haed
as an adze or used with a mallet lack objective critieria
with which to distinguish the correct one. e Hawaiian
terms for adzes do not clarify the situation or suggest that
a simple distinction is involved. Tools likely used as gouges
might have been called one of several names, such as pao,
ko‘i pāhoa, ko‘i hō‘oma, or ko‘i kahela (Holmes ; Pukui
and Elbert ), the distinctions among which are not
apparent in the literature and might have been lost. In
this situation, the archaeologist must distinguish gouges
from curved-edge adzes with characteristics inherent to
the stone tools themselves. How can these characteristics
be defined?
In practice, archaeologists typically distinguish gouges
from curved-edge adzes on the basis of size and shape.
Tools identified as gouges are typically small, narrow, and
elongated compared to other curved-edge adzes. ese
characteristics can be captured using weight as a proxy for
size and length index as a measure that combines informa-
tion on tool width and length (fig. ). e length index is
computed by multiplying the length of the tool by and
dividing the result by the width of the transverse section
at the shoulder (Garanger : ). A scatterplot of these
measures places small tools on the le, large tools on the
right, slender tools at the top, and stout tools at the bottom.
Plotted in this way, the nine tools typically identified as
Figure 3. Classicaon of curved-edge adzes and gouges. The dashed line is drawn as Length index = –50 + 100 * log(weight),
where weight is measured in grams. Tools that plot on or above the dashed line are gouges and those that plot below the
line are curved-edge adzes.
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
Oa−109
09511
B.01799
11018
B.00524
B.01695
B.02501
B.09360
C.08290
C.09210
D.00073
D.00435
1970.010.020
1970.286.017
1986.602
70.286.21
05853
10602
B.08629
C.09063
25204
70.286.22
10462
C.02493
200
300
400
500
600
3 4 5 6 7 8
Log weight (g)
Length index
Journal of Pacific Archaeology – Vol. · No. ·
gouges plot to the le and above the curved-edge tools
typically identified as adzes. As shown by the dashed line,
a gouge can be distinguished from a curved-edge adze by
a length index that is greater than or equal to – plus
times the natural logarithm of the weight in grams; the
length index of a curved-edge adze is less than this number.
As a group, the curved-edge adzes look different than
straight-edged adzes. In longitudinal plan, gouges have
narrow blades that oen taper to the cutting edge, and
many curved-edge adzes exhibit narrow shoulders and
wide edges. In contrast, the plans of straight-edge adzes are
more regular, with modest differences between the width
of the shoulder and the edge. is difference shows clearly
in a comparison of the distributions of cutting edge width
ratios for gouges, curved-edge adzes and straight-edge
adzes (fig. ). e cutting edge width ratio is designed to
distinguish among tools that expand or taper toward the
edge (Shipton et al. : ). It is defined as the width of
the edge divided by the maximum width at the shoulder,
where a ratio of indicates a tool with parallel sides, a ratio
greater than indicates a blade that expands toward the
edge, and a ratio less than indicates a blade that tapers to-
ward the edge. Here, the tendency of gouges to taper to the
edge shows in the mode at the lower tail of the distribution
for straight-edge adzes, and the tendency of the curved-
edge adzes to expand to the edge skews their distribution
to greater cutting edge width ratios that extend beyond the
upper range of straight-edge adze cutting edge width ratios.
e class of curved-edge adzes is rarely identified in
the literature on Hawaiian adzes, in large part because of a
preoccupation with the shape of the transverse section and
the presence or absence of a tang. e Hawaiian curved-
edge adzes are sometimes considered an early form with
a restricted temporal distribution, but direct evidence for
this claim is equivocal. An early description of Hawai-
ian adzes at Bishop Museum classifies them according to
whether a tang is present or absent and whether or not
the edge is wider than the poll (Brigham : ). It of-
fers no observations specific to the edge. Two small tools
with curved edges are described as gouges, one of which
is illustrated (Brigham : ). e gouges were found
suitable for carving wooden idols, ki‘i, because they fit ‘the
interior curve of the nostrils in two of the large idols in
this Museum’ (Brigham : ), and it was further noted
that the gouges might have been used with or without a
handle, one of them being too short to hold firmly in the
fingers. Stokes (: ) claims that gouges in Hawai‘i
were ‘used to perforate canoe sides for cord lashings.’ e
authoritative description of Hawaiian arts and cras (Buck
) makes only passing mention of adzes, presumably
because the work was unfinished at the time of the au-
thor’s death. Curved-edge adzes figure more prominently
in a culture historical study of East Polynesian adze rela-
tionships, where they are identified as hoofed adzes and
associated with plano-convex transverse sections (Emory
: –). Of the hoofed adzes identified at the time,
one was recovered from the ‘early levels’ of the K site at
Nu‘alolo Kai on Kaua‘i Island, leading to speculation that
it was an early form whose small numbers indicate a re-
stricted temporal distribution (Emory : ). A recent
estimate places the onset of cultural deposition at the K
site around (Hunt : ), several centuries
aer the islands were discovered by Polynesians (Athens
et al. ). A review of adzes recovered from Hawaiian ar-
chaeological sites distinguished Curve A edges as a curved
bevel form but did not identify Curve B edges (Cleghorn
: –). It found that adzes with a curved bevel form
are recovered from contexts that appear, on the basis of
Figure 4. Distribuon of cung edge width raos for 9 gouges, 15 curved-edge adzes, and 850 straight-edge complete adzes
at Bishop Museum.
0
1
2
3
4
0.5 1.0 1.5 2.0
Cutting edge width ratio
Density
Adze
curved−edge
gouge
straight−edge
Dye & Kahn – Functional Classification of Hawaiian Curved-Edge Adzes and Gouges
¹⁴C age determinations uncontrolled for in-built age, to be
relatively old. Although it might be tempting to view this
finding as supporting the hypothesis that the curved-edge
adze is an early form with a restricted temporal distri-
bution in Hawai‘i, subsequent experience re-dating and
re-analyzing several Hawaiian archaeological sites indi-
cates the pervasive effects of in-built age, with estimates
that are oen several hundred years too old (Bayman and
Dye : –). us, although the curved-edge adzes are
demonstrably rare, making up about percent of adzes
identified and described at Bishop Museum, their temporal
distribution is still poorly understood and it is premature
to identify them as an early form.
It is difficult to compare the Hawaiian curved-edge
adzes with other adze collections from Eastern Polynesia.
As noted earlier, the larger Hawaiian curved-edge adzes
resemble Māori adzes classified by Duff as Type . Never-
theless, the Duff Type adzes elsewhere in Eastern Poly-
nesia are functionally distinct from their Māori counter-
parts (Turner : ) and appear to exhibit straight edges,
rather than curved edges. Another difficulty is the habit of
grouping curved-edge gouges with straight-edged chisels
because of their similarly diminutive size. Nevertheless,
some rough comparisons are possible because in Eastern
Polynesia outside of Hawai‘i: i) the curved-edge adze is
associated with oval and plano-convex transverse sections,
which are oen identified; and ii) most authors treat the
smaller tools separately, albeit oen with some confusion
over the distinction between straight-edged chisels and
curved-edge gouges (Hiroa : ).
In New Zealand, curved-edge Type adzes make up
nearly percent of , adzes in museum collections
(Turner : ). In addition, some curved-edge adzes
were assigned to Duff’s Type , which includes chisels
and gouges, on the basis of functional and morphological
similarity (Turner : ). Although the morphological
similarity of chisels and gouges is striking, the claim for
functional similarity appears to have looked to the similar
ways chisels and gouges might be applied to the work, with
the motive force applied by striking the end of the tool
or its handle with a mallet. is definition of function as
the manner in which the tool was used makes classifica-
tion problematic because evidence for how the tool was
used is oen difficult or impossible to observe (Turner
: ). Also, it contrasts with a more restricted concept
of function as the intended result of applying the tool to
the work. If this restricted definition were used, then the
curved-edge tools assigned to Type would be assigned
to Type , instead.
Two curved-edge gouges from the Marquesas in Bish-
op Museum have been identified as examples of ‘imple-
ments uncommon to the group’ (Linton : ). Excava-
tions at the Ha‘atuatua site recovered nine gouges, which
were grouped as Ha‘e‘eka Type (Suggs : ). Five other
Ha‘e‘eka Type gouges were collected from surface contexts
in the Marquesas (Suggs : ). In addition, the Hatiheu
type, a rare find in the Marquesas, exhibits a bevel that is
‘somewhat concave’ (Suggs : ) and might have sup-
ported a curved edge, although the shape of the edge is
not described. Four basalt gouges were recovered from the
Hanamiai site on Tahuata (Rolett : –).
At Tubuai in the Austral Islands a ‘gouge with a
rounded edge and hollow ground bevel … belongs to a
type found intermittently throughout Polynesia’ (Stokes
: ). e Type adzes reported from Rurutu appear
to exhibit straight edges (Vérin ) and curved-edge
adzes are rare or absent in the collections made at the Peva
site (Bollt ).
Illustrations of several adzes from Easter Island sug-
gest curved edges. ree small implements identified as
chisels or gouges appear to have curved edges (Figueroa
and Sanchez Fig. a–c), as do two grooved butt adzes
(Figueroa and Sanchez Fig. a, b). In addition, adzes
classified as Type E, which is characterized by a plano-
convex transverse section, are reported as present on Easter
Island, the Marquesas, and possibly Pitcairn (Figueroa and
Sanchez : ).
e Huahine site in the Society Islands yielded eight
adzes with ‘(s)emicircular (quadrangular-oval and plano-
convex) [sections] with curved cutting edge’ (Emory :
).
Quantitative data on the relative proportion of plano-
convex adzes in Eastern Polynesian adze collections in-
dicate they make up: i) more than percent of a small
collection from excavations at a burial ground on Maupiti
in the Society Islands; ii) more than percent of a col-
lection made in the lagoon pass on Maupiti by divers; iii)
less than percent and slightly more than two percent of
other collections in the Society Islands; iv) slightly more
than percent of the adzes recovered during excavations
at the Hane site on Ua Huka in the Marquesas Islands; v)
more than percent of various museum and private col-
lections from the Marquesas; vi) more than percent of
adzes from Easter Island; and vii) less than percent of two
Hawaiian collections (Emory Tables –). ese rough
comparisons indicate that curved-edge tools are relatively
rare wherever they are found in Eastern Polynesia, and that
Hawai‘i is not unusual in this regard. ey make up less
than percent of most collections, the exceptions being
two small collections from the island of Maupiti and the
large collection of Māori adzes in New Zealand, where the
relative proportion of curved-edge adzes is greater than
percent, more than double what it is elsewhere in the
region.
Curved-edge adzes can be distinguished from i) straight-
edge adzes by a curved edge when viewed from the distal
end, and ii) axes by the presence of a single bevel. A curve
B edge and a single bevel are the necessary and sufficient
conditions for membership in the curved-edge adze class.
Journal of Pacific Archaeology – Vol. · No. ·
e primary role assigned to edge shape in this classifica-
tion reflects the importance of edge attributes to the func-
tion of the tool (Best ).
e curved-edge adze class can be divided into curved-
edge adzes and gouges on the basis of tool shape and size.
Gouges are small curved-edge adzes that are compara-
tively long and slender. ey can be unambiguously distin-
guished from curved-edge adzes using weight as a proxy
for size and length index as a proxy for shape (see fig.).
e curved-edge Hawaiian adzes identified in Bishop
Museum collections can be divided in two classes based on
weight (w) and length index (l) (see fig.). A curved-edge
adze is classified as a gouge when l ≥ – + * log(w)
and as a curved-edge adze when l < – + * log(w). Us-
ing this criterion, the Bishop Museum collection includes
nine gouges and curved-edge adzes. Archaeologists who
use these criteria to identify gouges and curve-edge adzes
should find the results of their efforts confidently repli-
cated by independent analysts.
From the point of view of an archaeological observer,
the function of a curved-edge adze is to shape a curved
surface when applied to the work, which is typically a piece
of wood. In practice a curved-edge adze might be applied
to the work in any number of ways, including direct per-
cussion, either with the adze haed to a handle or held di-
rectly in the hand, and indirect percussion, where a mallet
is used either to strike the poll of the adze directly or the
end of handle to which the adze has been haed in line.
Inferences about how a particular curved-edge adze was
used in practice might be based on observations of use
wear and the position of flake scars that are distinct from
the observations of edge curve, length, width, and weight
used to identify curved-edge adzes and gouges.
Acknowledgments
We would like to thank Betty Kam (former director of
the Bishop Museum Cultural Resources Division), Alice
Christophe (Ethnology Collections Manager), and their
assistants for providing access to the ethnographic col-
lections. Several Bishop Museum Archaeology Collection
Managers, in particular Charmaine Wong, likewise facili-
tated our research by providing access to the Hawaiian
archaeological collections. Diana Izdbeski is thanked for
draing the artifact illustrations. Two anonymous review-
ers offered several helpful suggestions on an earlier dra
of the paper.
References
Athens, J.S., Rieth, T.M. and Dye, T.S. (). A paleoenviron-
mental and archaeological model-based age estimate for
the colonization of Hawai‘i. American Antiquity : –.
Bayman, J.M. and Dye, T.S. (). Hawaii’s Past in a World of
Pacific Islands, Press, Washington, DC.
Best, S. (). e Maori adze: An explanation for change. Journal
of the Polynesian Society : –.
Blackwood, B. (). e Technology of a Modern Stone Age Peo-
ple in New Guinea, Oxford University Press, Oxford, UK.
Bollt, R. (). Peva: e Archaeology of an Austral Island Settle-
ment, Bishop Museum Press, Honolulu, HI.
Brigham, W.T. (). Stone Implements and Stone Work of the
Ancient Hawaiians, Bishop Museum Press, Honolulu, HI.
Buck, P.H. (). Arts and Cras of Hawaii, Bishop Museum
Press, Honolulu, HI.
Buck, P.H., Emory, K.P., Skinner, H.D. and Stokes, J.F.G. ().
Terminology for ground stone cutting-implements in Poly-
nesia. Journal of the Polynesian Society : –.
Clarkson, C., Shipton, C. and Weisler, M.I. (). Determining
the reduction sequence of Hawaiian quadrangular adzes
using D approaches: A case study from Moloka‘i. Journal
of Archaeological Science : –.
Clarkson, C., Shipton, C. and Weisler, M.I. (). Front, back and
sides: Experimental replication and archaeological analysis
of Hawaiian adzes and associated debitage. Archaeology in
Oceania : –.
Cleghorn, P.L. (). An historical review of Polynesian stone
adze studies. Journal of the Polynesian Society : –.
Cleghorn, P.L. (). A Hawaiian adze sequence or just differ-
ent kinds of adzes? New Zealand Journal of Archaeology
: –.
Cotterell, B. and Kamminga, J. (). Mechanics of Pre-Industrial
Technolog y, Cambridge University Press, Cambridge, UK.
Duff, R. (). Neolithic adzes of Eastern Polynesia. In J.D. Free-
man and W.R. Geddes (eds.), Anthropology in the South Seas:
Essays Presented to H.D. Skinner, omas Avery & Sons, New
Plymouth, NZ, pp. –.
Dunnell, R.C. (). Systematics in Prehistory, Free Press, New
York, NY.
Dunnell, R.C. (). Style and function: A fundamental dichot-
omy. American Antiquity : –.
Dunnell, R.C. (). Methodological issues in Americanist ar-
tifact classification. In M.B. Schiffer (ed.), Advances in Ar-
chaeological Method and eory, Academic Press, New York,
NY, pp.–.
Emory, K.P. (). East Polynesian relationships as revealed
through adzes. In I. Yawata and Y.H. Sinoto (eds.), Prehistoric
Culture in Oceania: A Symposium, Bishop Museum Press,
Honolulu, HI, pp. –.
Emory, K.P. (). e Societies. In J.D. Jennings (ed.), e Pre-
history of Polynesia, Harvard University Press, Cambridge,
MA, pp. –.
Figueroa, G. and Sanchez, E. (). Adzes of certain kinds from
Eastern Polynesia. In T. Heyerdahl and E.N. Ferdon (eds.),
Reports. Norwegian Archaeological Expedition to Easter Is-
land and the East Pacific, Forum, Stockholm, SE, pp. –.
Garanger, J. (). Herminettes lithiques Océaniennes: Éléments
de typologie. Journal de la Société des Océanistes : –.
Hiroa, T.R. (). Samoan Material Culture, B.P. Bishop Museum,
Honolulu, HI.
Holmes, T. (). e Hawaiian Canoe, Editions Limited, Hono-
lulu, HI.
Dye & Kahn – Functional Classification of Hawaiian Curved-Edge Adzes and Gouges
Hulswit, M. (). Teleology. In, e Commens Encyclopedia: e
Digital Encyclopedia of Peirce Studies. New Edition.
Hunt, T.L. (). Archaeological stratigraphy and chronology
at Nu‘alolo Kai, Nā Pali District, Kaua‘i. In M.T. Carson and
M.W. Graves (eds.), Na Mea Kahiko O Kaua‘i: Archaeological
Studies in Kaua‘i, Society for Hawaiian Archaeology, Hono-
lulu, HI, pp. –.
Kahn, J.G. and Dye, T.S. (). A note on Hawaiian stone axes.
Journal of Pacific Archaeology : –.
Kahn, J.G. and Dye, T.S. (). Supplementary Material for a
Functional Classification of Hawaiian Curved-Edge Adzes
and Gouges. https://doi.org/./kjp-p
Linton, R. (). e Material Culture of the Marquesas Islands,
Bishop Museum Press, Honolulu, HI.
Pukui, M.K. and Elbert, S.H. (). Hawaiian Dictionary, Fourth,
University of Hawaii Press, Honolulu, HI.
Richards, M.J. (). Duff’s adze typology versus Polynesian adze
exchanges. Archaeology in Oceania : –.
Rolett, B.V. (). Hanamiai: Prehistoric Colonization and
Cultural Change in the Marquesas Islands (East Polynesia),
Department of Anthropology; e Peabody Museum, Yale
University, New Haven, CT.
Searle, J.R. (). Social ontology and the philosophy of society.
In E. Margolis and S. Laurence (eds.), Creations of the Mind:
eories of Artifacts and eir Representation, Oxford Uni-
versity Press, Oxford, UK, pp. –.
Shipton, C., Weisler, M.I., Jacomb, C., Clarkson, C. and Walter,
R. (). A morphometric reassessment of Roger Duff’s
Polynesian adze typology. Journal of Archaeological Science:
Reports : –.
Sinton, J.M. and Sinoto, Y.H. (). Geochemical and petrograph-
ic evidence for adze production at Pu‘u Pāpa‘i, East Moloka‘i,
Hawai‘i. Journal of Archaeological Science: Reports : –.
Stokes, J.F.G. (). In, Ethnology of Tubuai, B.P. Bishop Museum,
Honolulu, HI, pp. –.
Suggs, R.C. (). e Archaeology of Nuku Hiva, Marquesas
Islands, French Polynesia, American Museum of Natural His-
tory, New York, NY.
omas, N., Adams, J., Lythberg, B., Nuku, M. and Salmond, A.
eds. (). Artefacts of encounter: Cook’s voyages, colonial
collecting and museum histories, University of Hawai‘i Press,
Honolulu, HI.
Turner, M. (). e function, design and distribution of New
Zealand adzes, PhD thesis, University of Auckland, Auck-
land, NZ.
Turner, M. (). Functional and technological explanations for
the variation among early New Zealand adzes. New Zealand
Journal of Archaeology : –.
Turner, M. and Bonica, D. (). Following the flake trail: Adze
production on the Coromandel east coast, New Zealand.
New Zealand Journal of Archaeology : –.
Vérin, P. (). L’Ancienne Civlisation de Rurutu (îles Australes,
Polynésie Française): La Période Classique, , Paris, FR.
