Petrological Sourcing of Polynesian Lithic Artifacts:Criteria, Methods, and Uncertainties

Society of Hawaiian ArcheologyEarth Day, 2015

John SintonDepartment of Geology and Geophysics

University of Hawai‘i at Mānoa

Petrology* of Volcanic Rocks*Petrology – the study of rocks, Πέτρος

Some things that can be measured in volcanic rocks:Texture

grainsize, fabric

Mineralogy

Chemical Composition

Major elements; trace elements; isotopic ratios

Age

Rock classification schemes combine chemical composition and texture

Basalt: a fine-grained, mafic volcanic rock, typically with <52.5 wt % SiO2, blah, blah ……

Magmatic Processes

1. Melting- The genetic imprint when the magma is “born”

2. Differentiation- How that magma changes composition as it rises, and

resides prior to eruption

3. Eruption (or not)- The cooling history of a magma affects the mineralogy and

texture of the rock

Melting and differentiation combine to control chemical composition (like genetics and environment combine to affect human personality)

A devilA born devilOn whose natureNurture can never strike-Prospero’s description of Caliban

Alkalic differentia

tion tre

nd

Tholeiitic differentiation trend

Chemical effects of melting and magmatic differentiation

melting

Names of Hawaiian volcanic rocks (based on composition)

Alkalic series

Tholeiitic series

• Because so many variables combine to control the composition and texture of a volcanic rock, there is a supposition that each particular lava flow or dike is likely to be (nearly) unique.

• Two flows from the same volcano might bear similarities in some characteristics, but probably not all.

extent of partial

melting

mantle source

composition

crustal storage, fractionation, and assimilation

intrus

ion

enviro

nment

erup

tion

rate

eruption conditions and duration

What to measure?Elemental abundance

Atom

ic n

umbe

r

Neutron number

What to measure?Isotopic ratios

Silicon and oxygen comprise ~45% (by weight) of the entire planet Earth

… and almost 75% of Earth’s crust

The most important rock forming minerals are silicates (containing silicon and oxygen)

O2- 43.19

Si4+ 22.78

Ti4+ 2.34

Al+3 7.23

Fe+2, +3 10.88

Mn+2 0.16

Mg+2 3.12

Ca+2 6.99

Na+ 2.13

K+ 0.91

P+5 0.23

Sum 99.97

All Fe as FeO

SiO2 48.73

TiO2 3.91

Al2O3 13.67

FeO* 13.99

MnO 0.20

MgO 5.17

CaO 9.78

Na2O 2.87

K2O 1.10

P2O5 0.54

Sum 99.97

All Fe as Fe2O3

SiO2 47.98

TiO2 3.85

Al2O3 13.46

Fe2O3* 15.31

MnO 0.20

MgO 5.09

CaO 9.63

Na2O 2.83

K2O 1.08

P2O5 0.53

Sum 99.97

Normalized

SiO2 48.00

TiO2 3.85

Al2O3 13.46

Fe2O3* 15.32

MnO 0.20

MgO 5.09

CaO 9.64

Na2O 2.83

K2O 1.08

P2O5 0.53

Sum 100.00

Major element* chemical analyses (all values in weight per cent)

*Elements that typically occur in abundances greater than 0.1 weight % and which together make up >99 % by weight of any given igneous rock

The same data expressed as oxide wt %

Note that analytical total is conserved

AbsolutePrecision

DetectionLimit

Sc 2 3

V 3 3

Cr 3 3

Co 3 3

Ni 3 3

Cu 3 3

Zn 3 3

Rb 1 0.6

Sr 1 1

Y 1 1

Zr 2 1

Nb 1 0.5

Ba 10 8

Pb 1 2

Th 1 2

U 1 2

La

Ce

Pr

Nd

Sm

Eu

Gd

Tb

Dy

Ho

Er

Tm

Yb

Lu

Hf

Ta

Trace elements by WD-XRF (ppm)

Additional trace elements analyzed well by ICP-MS

Typical absolute precision ~0.5 ppm

Typical LLD ~0.1 ppm

The confidence with which an artifact can be “sourced” to a specific location depends on two principal uncertainties

1. Number of elements (observations) and reliability of data obtained on the artifact

2. Composition and variability of the source region, as determined by number and quality of analytical data for that location

Each of these uncertainties can be quantified

Uncertainties associated with geochemical data1. Precision: How reproducible are the data? Will you get the same number if

you re-analyze a particular sample by the same method in the same lab?

2. Accuracy: How close are the analytical data to the absolute value in the sample?

Not shown: Pu‘u Wa‘awa‘a glass source

Most of the known quarries can be distinguished

Not many data for most Hawaiian quarries

Source Variability: What is the variation of the source? How well characterized is it?

Hawaiian Quarries: elements analyzed well by XRF

- Many quarries indistinguishable with these elements

- WD-XRF uncertainties smaller than symbols

- ED-XRF fields from Mintmeier et al. (2012)

Case Studies

1. Vitaria Quarry, Rurutu, Austral Islands [Rolett et al., JAS 2014]

Marquesas

Society Is.

Tuamotus

Austral Is.

Rurutu

RaivavaeTubuai

Tahiti

Gambier

Geologic map (Guille et al., 2000)

Adze quarry associated with major ceremonial complex

Viliamu Teuruarii[Rolett et al., JAS 2014]

The Vitaria source composition is well defined by geological samples and artifacts

most differentiated (highest SiO2, lowest MgO) of the Recent Volcanics on Rurutu

[Rolett et al., JAS 2014]

• Mean ± 1 standard deviation from the mean for each element or oxide

• The Vitaria source is relatively uniform in composition

Mean composition and variability of Vitaria source

(mean + 1s)/mean

(mean - 1s)/mean

[Rolett et al., JAS 2014]

Unknowns (with uncertainty) compared to source

• Unknowns compared to mean source composition.

• Error bars show analytical uncertainty for each element or oxide

• Some adzes from Raivavae match the Vitaria (Rurutu) source for all elements within uncertainty

Major element uncertainties smaller than symbol size.

Raivavae adzes from Peabody Museum

[Rolett et al., JAS 2014]

• Raivavae adze E36447 is not from the Vitaria source, (likely from Tubuai)

[Rolett et al., JAS 2014]

Unknowns (with uncertainty) compared to source

• Some elements (Al2O3, FeO, Zn, Nb) are consistent with Vitaria source, other elements are not

Case Studies2. Tuamotu adze (C7727)

Collerson & Weisler, Science, 2007

Tuamotu adze C7727 assigned a source on Kaho‘olawe, based on isotope ratios

C7727

A source in Samoa, Society Islands, Marquesas, and Austral Islands is not permitted by isotopic data

A source in the Hawaiian Islands is consistent with isotopic data

C7727

analytical uncertainty

Kaho‘olawe

Mō‘iwi

Adze C7727 is unlike any analyzed rock from Kaho‘olawe

2. Tuamotu adze C7727

Isotopic data consistent with a source in the Hawaiian Islands

Major and trace element data inconsistent with a source on Kaho‘olawe

Isotopic data alone can indicate or eliminate island chains; rarely can such data identify specific volcanoes

Case Studies

3. Unusual adze from ocean near Sand Island, O‘ahu

Measurements of 870 Hawaiian adzes in Bishop Museum collection (T. Dye & J. Kahn) http://tsdye.com/adzes/adzes.html

Sinton & Sinoto, in prep.

Unusual apatite-bearing alkalic lava, with flow layering

Indistinguishable from sample C-159 (Macdonald, 1968)

Distinctive- Grainsize- Flow layering- Brown apatite + magnetite

Pu‘u Pāpa‘iKeonekū‘ino, East Moloka‘i

Geology modified from Stearns and Macdonald, 1947)

Adze 1986.602 and Pu‘u Pāpa‘i share unusual chemical compositionNe-hawaiite with low SiO2, and extreme P2O5 (highest in Hawaiian Islands)

… and similar, equally unusual apatite compositions

Ca-phosphate: Ca5(PO4)3(OH, F, Cl)

1986.602 EMO-3 EMO-5CaO 52.94 53.14 53.99P2O5 40.06 40.38 40.82FeO 0.75 0.53 0.37MgO 0.49 0.47 0.27Na2O 0.24 0.26 0.22Ce2O3 0.16 0.17 0.18Cl 0.30 0.35 0.42F 1.76 1.67 1.70H2O* 0.89 0.91 0.89TOTAL 97.59 97.87 98.86

ions on basis of 25 oxygensCa 9.89 9.86 9.95P 5.91 5.92 5.94Fe 0.11 0.08 0.06Mg 0.13 0.12 0.07Na 0.08 0.09 0.07Ce 0.01 0.01 0.01Cl 0.09 0.10 0.12F 0.97 0.89 0.93H 1.03 1.06 1.02Sum 18.21 18.13 18.17

Data by UH microprobe; H2O calculated onbasis of stoichiometric OH+F+Cl

Working area at Pu‘u Pāpa‘i

3. Adze BP 1986.602

Adze rock indistinguishable from Pu‘u Pāpa‘i - texture- mineralogy- whole-rock composition- apatite compositions

This composition is extreme in P2O5 (and Sr) for the Hawaiian Islands and elsewhere in Polynesia

Likelyhood that Earth exactly repeated itself close to zero

Evidence for working of cores present in the field

Elem

ent Y

Element X

Case Studies

4. HypotheticalArtifact analyzed in lab A

Source rock analyzed in Lab B

It’s a (statistical) match! (but not a very good one)

What now?

1. Analyze artifact in Lab B

2. Analyze elements M, N, O, P, Q, R…….

3. Make a thin section

4. Consider other sources

“Sourcing” - matching analytical or observational data to that of known quarries, volcanic units, volcanoes, island groups ….

The confidence with which an artifact can be “sourced” to a specific location depends on two principal uncertainties

Composition and variability of the source region

Few quarries are presently adequately characterized

1. Quarry sampling depends on geology- dikes, lava flows, colluvial scatters and alluvial deposits will require

different sampling strategies

2. Analyze with the most comprehensive and best data available - petrography, mineralogy, texture- major element oxides- range of trace elements

1. No free lunchMore and better data are preferable to fewer and less precise data

2. One size does not fit allNested approachEmphasize diagnostic characteristics

The confidence with which an artifact can be “sourced” to a specific location depends on two principal uncertainties

How well analyzed is the artifact

Collaborators:Paul Cleghorn, Tom Dye, Steve Eminger, Jenny Kahn, Pat Kirch, Peter Mills, Barry Rolett, Yosi Sinoto, Marshall Weisler

MAHALO