Full text of "Marianas prehistory: archaeological survey and excavations on Saipan, Tinian and Rota"

See other formats

este

=s
=

t
HF Dot f
RS eee
He ati Rev pe bar Se
AT ite Ro ee iS
‘ ayaa eh Wey } .

~~ eS
PAS ee

Sate

ey ‘st
Hane Bu
panpel

—-.

Sata baatt ; |
i | hy *

foie tay
sein ggeas le
ret?

*
et is
-t Se
: Hee een a
= ut ney} Soars

vet ‘aie.

4 ce sd i

iis mere ht
eee — :

oe ee

“4th tat ie

re
“a
patie
4 +
Apts *

ied a
eee

ha _ tbe Bai
relat

(4)

HES
ke ce see

Phebe

+ +
id :

inte ‘ty! i
fer aE N i
Aiea Nats
aneuernt
Pa ae
wath ie rita ae
‘

bay sa
Hien
ae ark
oe Si uh

yea
ba Ae 4 re
iyo quit ep ae
oe! Me

ai a
ti ai ae hie

hi iis ese ms ns} sa
? RD iy pant

i)
Shi spete
.

Aa ak t
MG i rb Pe
rym a Wh re 3

tary

rete
sie ae

Pp aan Acnne wh, i
aah NR a aN
ait ee care
ha ete Si

aoe. £
Tbe
-

‘<

42823

$2a°3*.

eae! ie a i

4 ital hy : ea
ai estes
PAE Aa! plocgerse tie

Hehe
jg

Teta oer, Bi ata Ma tied sep
ts a attain

=
pee

6
—

pane i et

pale ts pe hh
eee is oy ee
es tate late

ee.
! ena

mM igh
feat oe

hie gece et
ce aie
pairs

BS Me

a
ae

= a ee
et ET Aww |

: fh

=e:
(

? C

MARIANAS PREHISTORY

ARCHAEOLOGICAL SURVEY AND EXCAVATIONS
On SAIPAN, TINIAN AND ROTA

ALEXANDER SPOEHR

d
XS
‘
4
; :
!
3
.
ie

FIELDIANA: ANTHROPOLOGY
VOLUME 48

Published by
CHICAGO NATURAL HISTORY MUSEUM

JUNE 24, 1957

FIELDIANA: ANTHROPOLOGY

A Continuation of the
ANTHROPOLOGICAL SERIES
of

FIELD MUSEUM OF NATURAL HISTORY

VOLUME 48

CHICAGO
NATURAL HISTORY
MUSEUM

CHICAGO NATURAL HISTORY MUSEUM
CHICAGO, U.S.A.
1957 THE LIBRARY OF THE

JUL 8 1957

UNIVERSITY OF ILLINOIS

Editors

PAUL S. MARTIN
Chief Curator, Department of Anthropology

LILLIAN A. ROSS
Associate Editor, Scientific Publications

MARIANAS PREHISTORY

ARCHAEOLOGICAL SURVEY AND EXCAVATIONS

On SAIPAN, TINIAN AND ROTA

7
-
-

moet
oa 1
on

tc
inn ee :
ak ee

ioe

ie
nr
oe

De ae ;
ae ee
oe oe a,
Ce
ie
on

MARIANAS PREHISTORY

ARCHAEOLOGICAL SURVEY AND EXCAVATIONS

On SAIPAN, TINIAN AND ROTA

ALEXANDER SPOEHR
Formerly Curator, Oceanic Ethnology

Director, Bernice P. Bishop Museum
Professor of Anthropology, Yale University

FIELDIANA: ANTHROPOLOGY
VOLUME 48

Published by
CHICAGO NATURAL HISTORY MUSEUM

JUNE 24, 1957

PRINTED WITH THE ASSISTANCE OF
The Edward E. Ayer Lecture Foundation Fund

Library of Congress catalog card number: 57—12368

PRINTED IN THE UNITED STATES OF AMERICA
BY CHICAGO NATURAL HISTORY MUSEUM PRESS

Preface

This is the final report on anthropological field research conducted for
Chicago Natural History Museum in the Mariana Islands of Micronesia
during 1949-50. An earlier monograph published by the Museum covered
the ethnological results of the field work, and was primarily a historical
analysis of culture change in the Marianas from the time of first contact
with Europeans in the sixteenth century up to contemporary times
(Spoehr, 1954). The present report deals with the archaeology of the
Marianas in an attempt at extending the historical record backward in
time from the point of early European contact. The two reports therefore
complement each other.

Financial support and sponsorship of the Marianas research was pro-
vided by Chicago Natural History Museum. I am deeply indebted to the
Museum for the generous support it has always accorded me. In connec-
tion with the Marianas research, I particularly wish to record my apprecia-
tion to Mr. Stanley Field, President of the Museum, Dr. Clifford C. Gregg,
Director, and Dr. Paul S. Martin, Chief Curator of the Department of
Anthropology. Miss Agnes McNary, formerly Departmental Secretary, has
assisted with many details, and Mr. Phillip Lewis has given assistance in
the final drafting of ground plans. For identification of bone, shell, and
stone specimens, I am indebted to Mrs. Dorothy B. Foss, formerly Assistant
in Anatomy and Osteology, Dr. Fritz Haas, Curator, Lower Invertebrates,
and Dr. Sharat K. Roy, Chief Curator of the Department of Geology. Miss
Lillian Ross, Associate Editor of Scientific Publications, has extended her
generous aid in seeing this report through the press.

The Museum’s work in the Marianas received also the sponsorship of
the Pacific Science Board, National Academy of Sciences—National Re-
search Council. The Board rendered invaluable assistance in facilitating
transportation and field work, and I am particularly grateful to Mr.
Harold J. Coolidge, Executive Director of the Board, for his personal in-
terest in the success of the work. Miss Ernestine Akers, of the Honolulu

_ office of the Board, was very helpful in expediting numerous logistic details
in co-operation with the Navy.

Without the generous co-operation of the Department of the Navy, the
Museum’s field project in the Marianas would not have been possible.

8 PREFACE

Special thanks are accorded to Admiral A. W. Radford, Rear Admiral
L. S. Fiske, Rear Admiral E. C. Ewen, Captain W. C. Holt, Captain S. P.
Sanford, Commander J. R. Grey, Commander W. R. Lowndes, Com-
mander E. C. Powell, Lieutenant-Commander J. B. Johnson, Lieutenant
C. J. Carey, Lieutenant J. S. Broadbent, Lieutenant M. S. Bowman,
Lieutenant R. F. Roche, Lieutenant B. I. Rosser, Lieutenant J. W. Millar,
Lieutenant D. D. Moore, Lieutenant (jg) E. H. Gummerson, Lieutenant
(jg) S. Weinstein, Lieutenant (jg) W. E. Laskowski, Lieutenant (jg) R. K.
Hoffman, Chief C. T. Smallwood, Chief J. T. McCrosson, and Staff
Sergeant J. E. Hinkle, U.S.M.C. Among civilian specialists attached to the
Saipan District, I am particularly indebted to Messrs. Frank L. Brown,
Cyrus F. Quick, Kan Akatani, and Ernest G. Holt. Grateful acknowledg-
ment is made for the assistance and hospitality freely offered by all these
personnel and by their families. I particularly wish to express my thanks
to Mrs. W. C. Holt for her aid in the archaeological work.

The work on Saipan was greatly aided by the friendly interest of many
Chamorro friends to whom I wish to express my appreciation: Messrs.
Ignacio V. Benavente, Vicente de L. Guerrero, Ricardo T. Borja, Elias P.
Sablan, William S. Reyes, Juan M. Ada, Antonio Cabrera, Joaquin
Cabrera, and Ramon Borja. On Tinian, I am particularly indebted to the
survey team of the United States Geological Survey then operating on
Tinian, and consisting of Messrs. Harold W. Burke, David B. Doan, and
Harold G. May. The Geological Survey assisted tremendously in the loca-
tion of sites, in the loan of surveying equipment, and in extending the al-
ways generous hospitality of their quarters. For friendly co-operation on
Tinian I am also indebted to the Reverend Marchand Pellet, OFM, Cap.,
and to Messrs. Henry Fleming and Henry Hoffschneider.

Digging under a tropical sun is hard and tedious work. Yet I have never
worked with a more effective and conscientious digging crew than during
the Museum’s work on Saipan and Tinian. My special thanks are accorded
to Benedicto Taisacan, Vicente Selepeo, Joaquin Palacios, José Palacios,
Henrico Lisama, Serafin Atalig, Pablo Piteg, and Ignacio Sablan.

To Dr. Paul Fejos and the Wenner-Gren Foundation for Anthropologi-
cal Research, Inc., grateful acknowledgment is made for the loan of photo-
graphic equipment.

The results of the field work have been greatly facilitated through valu-
able assistance rendered by Dr. Preston E. Cloud, Jr., of the United States
Geological Survey, on the geological history of the Chalan Piao area on
Saipan; and by Dr. T. Dale Stewart, of the United States National Mu-
seum, on the paleopathology of prehistoric skeletal material.

Since assuming my duties at the Bernice P. Bishop Museum, I have been
fortunate in being able to examine the manuscript field notes of Hans G.

PREFACE 9

Hornbostel on the Marianas. For assistance rendered at Bishop Museum, I
am indebted to Miss Margaret Titcomb, Librarian, who made available
the Hornbostel notes; and to Miss Mary Stacey, Secretary to the Director,
who typed the manuscript of this report.

Lastly, my greatest debt is to my wife, Anne Harding Spoehr. She
handled virtually all the logistics on Saipan, directed the work of cleaning
the thousands of potsherds and artifacts found in the field work, and gave
invaluable encouragement in the completion of this report, despite the
many interruptions that have occurred since the termination of the field
work. I am indeed grateful to her.

ALEXANDER SPOEHR
December 1, 1955

[* a Aa
es

oe
os

re,

ae
ae
ae

va)
Se el
ie

Aa pe cn
Pea
: 7

i”

IT.

10 i

TV.

Contents

List OF ILLUSTRATIONS .

. INTRODUCTION
Problems of Marais: Bichistiry
The Mariana Islands : :
Chamorro Culture at the Time of Barly Eupvesn Contact
Selection of Islands for Archaeological Field Work
Procedures
SURVEY AND EXCAVATIONS ON SAIPAN.

Physical Characteristics of Saipan .
Survey

Excavations at Onan

Excavations at Laulau

Excavations at Oleai and Chalan Kija.
Excavations at Chalan Piao .

SURVEY AND EXCAVATIONS ON TINIAN

Physical Characteristics of Tinian
Survey

Excavations at the Blue Site
Excavations at the Taga Site
The Taga Quarry

ARCHAEOLOGICAL SURVEY OF ROTA

V. POTTERY .

Methods of Bacavation and kenalvats
Pottery Types
Marianas Plain
Marianas Fine-Line Incised
Marianas Cord-Marked
Marianas Trailed .
Marianas Red... ,
Lime-filled, Impressed “Trade Ware :

Local Differences and Temporal Relauonships evealed By Poery

Pottery and Baked Clay Artifacts

VI. Srone ARTIFACTS

Adzes .

Chisels .

Gouges and Axes
Slingstones
Pestles .

PAGE
13

17
20
22:
25
26
26

28
28
29
34
43
58
60

68
68
68
Ne
85
98

102

107
107
108
109
112
114
116
0G
120
22
129

131
131
136
137
137
140

12 CONTENTS

PAGE

IMiOrtAaTS: sae tins es ee ee he eres ca Say eh, eee ee ent ct lA

Mail sinter eee Cw an Oe oe ke ante we aay a ELAS
FlammerstGnes” 25 (we oy oe ah ee ae ce eh ee we 43

Sinkers eS ek ne er ee ae ew mY
Pendants 4ya oy ae oe Se a ed ee ee ea
Scrapersh 230 / Pe utr ee eo ee Seek, We eae ee ce ety Ree AD ee LY,

Knives BE ghey he Pore RTE ee era ee ie er a ea red ee 2 1
Unworked Cores, Flakes, and Stone Fragments . . . . . . . 150

VII. SHELL, BonE, AND METAL ARTIFACTS lane ee uae Uae ny oe Renee Geel
Shell Artifacts SIE oy Ar eet gg tee pt eres ee eae er te TOL

Bone Artifacts ee ey ey ee A ae a eee tee ee aks dN SEA Ko} 0)

Metali Artifacts? 4.82.) ett ane th en ee oe Nee, Ae Pace lO

VTS GONCEUSION wae eat Lieve ania fecal hak Way atin tacos Ua eed Sat Ae ee ae aC OS
Cultural:Chronology insthe:Miarianasi< ("2 Gite an ee OG
EatiesPhase7ot. Marianass@ultures oa 0 ot. 6, Geen ee i Ba eee Slt
Pre-Catte- Phase, Gultures sito, <1or ee ae ve eee eee ee Gu
@ultural-A fiiliationsiof the: Marianas: 3) (25) 3 a ee ey cee lS:
Refiectionsione Oceanic: brehistoryace 45) su sae eo eee anes ees
BIBLIGGRAPHY: “ever tary ae ee ee oe rr ei SO Ae oe Pa oe OS

ENDE XG,” ors Foes Mag nes see Se Ne ee ee eee ee ae ane G

OoOmWON HWDWN WO ND

—_
-_= ©

Ne ee ee ee
NO" 60) oI Os OW, pate OO SND

OOO oe Oona hOm No (DCN Ror rN iby Doni
wn OO WN A NN hh WH NHN = ©

List of ImIlustrations

TEXT FIGURES

. Micronesia, showing known limits of pottery distribution

. The Taga /atte on Tinian, as seen by Lord Anson in 1742

. Stone shaft and fallen capstone from a Jatte at Bapot, Saipan

. Map of the Mariana Islands

. Principal archaeological sites of Saipan rae
. Magicienne Bay, Saipan, showing location of Laulau and Bepees Sites
. Fafunchulujan Bay, Saipan, showing location of Fafiunchulujan Site
. Plan of latte and excavations at Objan, Saipan

. Upper: Objan Jatte cleared of vegetation. Lower: shaft 5

. View from above of excavations at Objan

. Shaft 7, Objan, Saipan .

. Soil sections, Objan, Saipan

. Plan of Laulau Site, Saipan j

. Plan of latte and excavations, House ry Paige Site, Spas

. House A, Laulau, Saipan .

. House A, Laulau, Saipan

. Soil section, House A, Laulau Site, Saipan

. Plan of excavations, Laulau Rock Shelter, Saipan

. Laulau Rock Shelter, Saipan .

. Soil section, Laulau Rock Shelter, Saipan

. Burial no. 6, Laulau Rock Shelter, Saipan

. Plan of excavations at Oleai Site, Saipan

. Oleai Site, Saipan ue

. Plan of excavations at Chalan Kija Site, Sieh

. Plan of excavations at Chalan Piao Site, Saipan

. Excavations at Chalan Piao Site, Saipan . Be ae ou:
. Oyster shell (Ostrea cucullata Lamarck) from Chalan Piao, Sane :
. Principal archaeological areas and sites of Tinian =e
. Location of Blue Site, House of Taga, and Taga Quarry, Tinian .
. Map of Blue Site, Tinian

. Plan of Blue I, Blue Site, Tinian

. Section through shafts 4 and 9, Blue I, Blue Site, Tinian

. Blue I, Blue Site, Tinian

PAGE
19

21
22
23
30
31
32
33
37
38
39
41
44
47
48
49
50
51
53
55
56
59
61
62
63
64
65
69
75

ats

79
80
81

34.
Sy
50:
Bc
38.
39:
40.
41.
42.
43.
44,
45,

46.
47.
48.
49.
50.

51.
52,

D5
54.

55.
56.
Bie

58.

oy,

60.

61.

62.
63.
64.
65.
66.
67.
68.
69,
aA;
Be:

LIST OF ILLUSTRATIONS

Shaft 7, Blue I, Blue Site, Tinian

Burials, Blue I, Blue Site, Tinian. Upper: Burial 2. Lower: Burial 1
Map of Taga Site as it was in 1924 (after sketch by Hornbostel)

Plan of House of Taga and of test trenches, Tinian

Shafts 4 and 9, House of Taga, Tinian .

Fallen shafts and capstones, House of Taga, Tinian as ke ate
Sections of shafts, capstones, and test trenches, House of Taga, Tinian .
Plan of Taga Quarry, Tinian

Quarry holes, Taga Quarry, Tinian . ae AE ie Gy oar ae ee
Map of Rota, showing principal areas of /atte sites and As Nieves Quarry
Plan of quarry at As Nieves, Rota (after Hornbostel) .

Sections through Capstones 1 and 5, As Nieves ee: Rota ane Horn-
bostel) Tag eae SL eee nabiae 2

Marianas Plain sherds
Graph showing thickness of Marianas Plain oe sherds .
Marianas Plain bowl

Marianas Plain rim sections

Upper row: Marianas Plain rim sherds (Type C) with finger impressions
around lip. Lower row: Marianas Fine-Line Incised sherds

Marianas Cord-Marked sherds

Marianas Trailed sherds

Marianas Red sherds _ .

Graph showing thickness of Marianas Red Body sherds

Marianas Red sherd sections .

Lime-filled, impressed sherds .

Rim sections of lime-filled, impressed sherds

Graph showing gross differences among sites in frequency of pottery types .

Graph showing percentage differences in one? ph aads ie ea Chalan Piao.
Lower: Laulau Rock Shelter

Graph showing gross differences among excavated sites in relative eau
cy of Marianas Plain rim types

Graph showing percentage differences among excavated sites in relative fre-
quency of Marianas Plain rim types gots

Pottery and baked clay artifacts .

Stone adzes, Type 1

Stone adzes, Type 2

Stone adzes, Type 3

Stone chisels

Slingstones with pointed ends

Slingstones with rounded ends

70. Pestles

Stone mortar found on Tinian near the House of Taga

Stone mortar in situ, Laulau House A, Saipan .

100
103
104

105
109
111
112
113

114
115
116
aly
118
119
121
121
122

123

125

125
130
132
134
135
137
139
139
141
143
143

74,
TDs
76.
das
78.
TEE
80.
81.
82.
83.
84.
85.

86.

87.
88.

89.

LIST OF ILLUSTRATIONS

Upper left: pears: sees Bent: maul. Center and lowest specimens: hammer-
stones : : : :

Hemispherical ee viewed from above

Pyramidal sinkers with rounded surfaces

Sinkers .

Pendants or sinkers

Scrapers te ee Pe ar oma ees ae
Upper three specimens: knives(?). Lower specimen: ene knife or ie
Unworked cores and flakes

Shell adzes and gouges

Pedalion shell knives or scrapers (?)

Turbo shell scrapers

Shell container for lime chewed with betel-nut

Top row: shell fish hooks. Center row: shell fish Sore Bottom: partially
completed gorge

Top row: shell beads(?). Center row: shell eecadanel). Bottom left: shell
bracelet. Bottom ete i san of Babirussa tusk, probably used as
ornament : ant cane 8, Core ete ar ek he ne oe Bs

Bone awls or eictine ee

Upper two specimens: fragments of spear points cut from human bone. Third
specimen from top: bone fish spear point(?). Bottom specimen: partially
sawed section of human long bone St Se eet or ek a

Metal artifacts .

PAGE

144
145
145
147
147
148
148
149
153
155
156
156

159

159
161

165
166

a
. ey

a ae

met

I. Introduction

The reconstruction of Oceanic prehistory has been based primarily on
the evidence of linguistics, ethnography, and physical anthropology, drawn
from the study of modern island peoples. In these fields of anthropology,
much work of major significance for Oceanic culture history has been ac-
complished and much remains to be done. Yet if we are ever to understand
the migrations of man into Oceania and the course of events that lie in the
Oceanic past, recourse must be had to archaeology. Archaeology can pro-
vide kinds of data that other disciplines cannot furnish. It is very largely on
the data of archaeology that future knowledge of Oceanic culture history
must be built.

Since World War II, there has been a freshening of interest in the
archaeology of the Pacific islands. Duff in New Zealand, Emory in Hawaii,
Gifford in Fiji, New Caledonia, and Yap, and Osborne in the Palaus have
produced, through their excavations and reports, valuable new perspec-
tives where none existed before. This post-war work is characterized by the
application of stratigraphic techniques to archaeological deposits. Oceanic
archaeology is finally getting down to the business of intensive digging in
an effort to determine sequences of culture change. The emphasis on dig-
ging, rather than only on description of surface remains, is a healthy trend
and is a forerunner of solid results in the study of Oceanic prehistory.

A second reason for the productivity of recent work and for an optimistic
prognosis for the fruitfulness of future field research is the development of
radiocarbon dating (Libby, 1951). Oceanic archaeology is handicapped
by a paucity of time indicators among the artifacts with which it deals. The
materials for constructing even a relative chronology of change are not
elaborate. Granted the problems surrounding radiocarbon dating, it is a
revolutionary development for which Oceanic archaeologists can give fer-
vent thanks each time they thrust a shovel into a Pacific island midden.

Of the three major regions of island Oceania—Polynesia, Micronesia,
and Melanesia—only Polynesia can boast of a series of professional reports
covering archaeological exploration that provides a basis for intensive ex-
cavation and field research. The pre-World War II literature on the
archaeology of Melanesia and Micronesia is, with a few exceptions, frag-
mentary and sketchy. The prehistory of Melanesia and Micronesia is a
truly uncultivated field of inquiry.

slg/

18 MARIANAS PREHISTORY

In contrast with Polynesia, however, both Melanesia and Micronesia
possess a most important aid to the archaeologist. This aid is the presence of
pottery, a fundamental indicator of culture change. The distribution of
pottery in Melanesia and Micronesia is an important element in planning
the strategy of field research in both areas. It would seem that one of the
best approaches to archaeological work in Melanesia and in Micronesia is
made by beginning at the margins of the pottery area and proceeding
westward, with the aim of cross-dating pottery sequences from individual
island groups. This is the strategy that Gifford has followed south of the
equator. Commencing with his important excavations on Fiji (Gifford,
1951), he and Shutler then pushed westward to New Caledonia, where
they have recently completed a second series of field investigations. This
same consideration of commencing at the margins of the pottery area, with
the hope of extending future work to the more complex areas to the west,
was the controlling factor in the selection of the Marianas for my own
archaeological field work in Micronesia.

The known limit of pottery distribution in Micronesia is shown in figure
1. Only the western part of Micronesia falls within the pottery area. Most
of Micronesia is excluded and together with Polynesia forms a great island
expanse where the prehistorian must use indicators other than pottery to
reconstruct sequences of culture change. The map shows that the Micro-
nesian pottery area is comprised by a southwest to northeast line of high
islands, with the Marianas at the outer end. In geographic terms, the
Marianas are truly marginal and very much at the end of the line. The
Palau Islands, on the other hand, lie directly in the path of presumed mi-
grations from Malaysia and probably formed a funnel through which man
passed into the Carolines, as well as into the Marianas. The Palaus ob-
viously hold the key to much of Micronesian prehistory.

Another important consideration in the selection of the Marianas for
field work was that it is the only major island group in Micronesia whose
archaeology has received much attention. During the 1920’s, H. G. Horn-
bostel, an observant collector, explored archaeological sites on Guam,
Rota, Tinian, and Saipan, made careful notes, collected specimens, and
conducted limited excavations. Although his work yielded no published
stratigraphy and was primarily confined to surface observations, his mate-
rials have proved very useful. Hornbostel’s collections and a major part of
his notes have been written up by Thompson (1932), who, in addition, has
provided an outline of Marianas ethno-history (1945). These reports,
taken in conjunction with the notes of other investigators, such as Safford
(1905) and Fritz (1904), provide a valuable starting point for more inten-
sive archaeological investigations.

‘uonnqiystp Aiajyjod jo sy] UMOUY SuIMmoys ‘eIsouoIIIPY “| “OLY

| aaa ee baz
: eOL! 209! oie
oe 5 ee
Vv s =
i ° $
ws rs YOLVNOF
'
at eon)
o .
1: Saul
+. “awene. <3 ON,
<a ri ere he ae E
2 3 3dVNOd ~ yn * ;
Sree ae
“. 3 oe eve > oe
i— Ol cot > .
: Sy
’ ae ; vy s
~
>
e °
. ee SONV1SI
ae sce? ANIddITIKd
a (2 ’
Sipe 6 7
eet ee ee .
+ 24VM Le Waar) ‘ A
-— 02 ca ed . ‘ ical
S !
\ ‘ F
. VSOWHOS
. . ‘ fe a» rik
.
Ola i x¢ ) ne ee
eALI 009) eOG!l NINOS z 00! oO! &: mR
| | | | ba we

20 MARIANAS PREHISTORY

PROBLEMS OF MARIANAS PREHISTORY

Unlike work done in areas of long-standing archaeological interest, such
as the American Southwest, it was impossible at the onset of the Chicago
Museum’s Marianas work to define specific problems of space and
time relationships as a guide to field operations. There were only the very
broad problems of which all Oceanic students are aware, such as the ques-
tion of prehistoric relationships of the Marianas with the Palaus and the
Philippines. Without a satisfactory basis of knowledge derived from pre-
vious archaeological work, however, such problems are too general to be
useful as controls for field work. The picture in the Marianas was that
facing any archaeologist entering an area where the only knowledge avail-
able is description of surface finds, plus a sketchy outline of the indigenous
culture at the time of contact.

Faced with this situation, I had two principal objectives in mind. The
first was to obtain, through the application of stratigraphic techniques, as
long a prehistoric culture sequence in the Marianas as was possible.
Linked with this objective was that of obtaining suitable materials for
radiocarbon dating, in order to tie to absolute dates any relative chronol-
ogy that could be determined.

The second objective was the more complete delineation of what I have
chosen to call the Latte Phase of Marianas culture. The most distinctive
archaeological sites in the Marianas consist of double rows of stone col-
umns, usually in sets of eight or ten, locally called latte. Early documentary
sources and the observations of Fritz, Safford, Hornbostel and others made
it certain that these columns were houseposts. The sets of stone columns
had, in places, been found in end-to-end rows, indicating a village pattern,
and around them were invariably discovered potsherds, artifacts, and the
debris of daily living. When the earliest explorers discovered these latte type
structures they were still in use. The antiquity of the earliest latte was
merely a matter of conjecture. A principal objective of the Marianas expe-
dition was to determine the time span of the Latte Phase, including both
architectural features and associated artifacts, and through excavation and
analysis of remains to clarify its diagnostic characteristics.

These were modest objectives. Yet in a new region of investigation, first
things must come first. In Micronesia the immediate need is a framework
of established spatial and temporal cultural relationships in prehistoric
times within and among major island groups. Until such a framework has
been established we shall advance but little in our understanding of the
prehistoric flow of events in Micronesia. It is my hope that the present
report will contribute toward satisfying this fundamental need.

‘TpLI Ul UosuY pioyy Aq Ud—S se ‘URIUT], UO 477) BBET 2YT, “7 “OILY

22 MARIANAS PREHISTORY

2 TOP VIEW
CJ
O

Present Ground Level

Old Ground Level

Y Y

SIDE VIEW
SO ele ee
So as ee a
FEET

Fic. 3. Stone shaft and fallen capstone from a /atte at Bapot, Saipan.

THE MARIANA ISLANDS

The Marianas consist of fifteen islands, lying in a long, flat arc from
Farallon de Pajaros (Uracas) in the north to Guam in the south, a distance
along the arc of nearly 500 miles (fig. 4). This chain can be divided into a
northern and a southern group. The islands of the northern group are
small, volcanic peaks rising abruptly from the sea. Those of the southern
group, consisting of Guam, Rota, Aguijan, Tinian, and Saipan, are larger
and are composed primarily of superimposed, coral limestone terraces rest-
ing on volcanic bases. The southern group of islands contains much greater
areas of level or gently sloping land and is much more suitable for human
occupation. The southern islands have probably always been the seat of
most of the population of the Marianas.

The combined land area of the Marianas is approximately 400 square
miles, but of this total, Guam accounts for more than 215 square miles.
The land areas and peak elevations of the individual islands are given in
the table on page 24.

tad
*FARALLON DE PAJAROS

147°

20° *MAUG

* ASUNCION

@ AGRIHAN

# PAGAN

20°

* ALAMAGAN
* GUGUAN

* SARIGUAN

@ ANATAHAN

18°

16 °———— FARALLON DE MEDINILLA °

E saipan
TINIAN $

* AGUIJAN

“P ROTA

16°

14°

GUAM
|
145°

147°

Fic. 4. Map of the Mariana Islands.
23

24 MARIANAS PREHISTORY

THE MARIANA ISLANDS!"

Area Peak Elevation
Name (sq. miles) (feet above sea level)
Farallon de Pajaros (Uracas)......... 0.79 1,047
Plt eet rete Neen Or eae 0.81 748
ASUNCION: aver eee Ga Re NORTEL Se EL ALLOD 23929
ABMBA c2icr ache nackte Aen eee ten Se ete 18.29 3,166
DAR ATI Re Se opie PrP sae ee Aen eee: ee ahs 18.65 1,883
Alamagan® i.2 cs Gib Bs Maes dogs eco 4.35 2,441
CUP MAN aise: Bk is cto een cess 1.61 988
Ahi gC UaNs. Coates ones, or aes etc 1395 1,801
Anatahanijs ataventco aes ianianeutek es 12.48 2,585
FarallonidesMedinilla: t4 35.2.0 wane 0.35 266
Salpansstod.ssrar are acanniep canine ase, cee aeeee 46.58 1,554
SN Sn bona Voge iy seen nen er nv eS (eh era ne 39.29 564
AGU aNy eer ene serenity ea EERIE Qeilah 584
ROAR Es A Om er eine ree ES OO) 1,612
(Gualtiverde ne pieces ere eter ete 215250 1,334
Total oasis connie omen 399.12

1 Data taken from Bryan (1946).

The Marianas lie both in the trade wind latitudes and on the eastern
fringe of the Asiatic monsoon area. The climatic conditions are thus largely
the product of the interplay of trade winds and monsoons. The trade winds
prevail through the first part of the year. In late spring or early summer
the winds shift to the south and southwest, but by the end of the summer
months the trade winds gradually begin to assume control once more.
Temperatures range from 75° to 85° F. Rainfall is abundant, with an an-
nual average on Saipan of approximately 82 inches. There are distinct wet
and dry seasons, the latter extending from about June to December, al-
though the onset of each season is not sharply marked. Also, periodic rains
can be expected during the dry season. Typhoons may occur at any time
but tend to fall in the period between August and November.

The vegetation of the islands, particularly the southern group, has been
so altered by human occupation that its original character is preserved
only in a few restricted areas, such as along cliffs and on steep slopes. This is
especially true of Saipan and Tinian, where most of the arable land was
brought under cultivation by the Japanese, to produce sugar cane. During
World War II, both Saipan and Tinian were invaded by American forces
and thereafter transformed into huge wartime bases. The military con-
struction attendant upon this transformation further altered the landscape.
By 1950 these installations had largely been abandoned, giving both Sai-
pan and Tinian a pronounced feeling of desolation, relieved only by local
areas that still retained their natural beauty.

INTRODUCTION 25

CHAMORRO CULTURE AT THE TIME OF EARLY
EUROPEAN CONTACT

At the time of first discovery and exploration by Europeans, the Mari-
anas were inhabited by a people speaking a single language and possessing
a homogeneous culture. The inhabitants of the islands came to be known
as Chamorros, the name that is still applied to them. Their history, follow-
ing their discovery and conquest by Spain, has been the subject of a con-
siderable literature. Thompson (1947), Joseph and Murray (1951), and
Spoehr (1954) give outlines of post-contact Chamorro history.

The first European to sight the Marianas was Magellan, in 1521. There-
after there were sporadic visits by European, principally Spanish, ships,
until 1668, at which time Spain formally entered upon an attempt to mis-
sionize and conquer the islands. By the close of the seventeenth century, the
Spanish conquest had been completed and all Chamorros removed to
Guam, except for a handful remaining on Rota.

The Chamorros of these early days were fishermen and farmers. They
were expert outrigger-canoe builders and skilled as sailors. In fishing, they
used nets, spears, and hooks and lines; they also gathered shellfish from the
reefs. As farmers, they raised yams, taro, bananas, breadfruit, sugar cane,
and coconut palms. In addition, they appear to have grown rice. Domesti-
cated animals, however, seem to have been conspicuously absent; the early
accounts contain no certain evidence of the dog or pig, and whether the
Chamorros possessed domesticated fowl is open to question.

Chamorro tools were of stone, shell, bone, and wood, for metal was
unknown. After the first contact with European ships, iron was eagerly
sought after. Pottery was extensively manufactured.

The Chamorros lived in small hamlets and villages, usually located
along the coast, although fertile interior areas were also occupied. The
houses were constructed of wood and thatch and some of them were erected
on stone houseposts. The bachelors’ house, in which were kept the skulls of
ancestors, was an established feature of the village plan.

The social organization is but sketchily known. A rigid stratification of
nobility and commoners prevailed. Some sort of maternal descent grouping
existed, though its exact nature is unknown. The relatively large size of
many latte house sites suggests the former presence of extended families.
In political organization, each island was divided into districts, but these
were not consolidated into large, well-knit political units. Warfare between
districts seems to have been the rule. In war, slings and spears were the
offensive weapons used; defensive measures included camouflaged pits
lined with sharpened bamboo stakes, a distinctively Malaysian feature.

26 MARIANAS PREHISTORY

Warfare was conducted primarily by stealth and ambush and was not
subject to a high degree of organization.

Further notes on Chamorro ethno-history are given in Thompson (1945)
and Spoehr (1954).

SELECTION OF ISLANDS FOR ARCHAEOLOGICAL
FIELD WORK

The objectives of the Museum’s 1949-50 expedition to the Marianas in-
cluded an ethnological study of Saipan as well as archaeological work.
For this reason, headquarters were maintained on Saipan. From an archae-
ological point of view, Guam had been better explored than either Saipan
or Tinian, so that an understanding of Marianas prehistory demanded
fuller coverage of the latter two islands. I decided, therefore, to concentrate
attention on Saipan and Tinian. From the map (fig. 4), these two islands
can be seen to fall together as a spatial unit, in contrast with the southerly
pair of islands, Guam and Rota. There is consequently good reason to con-
sider Saipan and Tinian together. After the completion of the work on
Saipan and Tinian, I conducted a brief reconnaissance of Rota to deter-
mine in a preliminary way the extent and nature of the remains there.

PROCEDURES

An archaeological survey was commenced on Saipan on November 18,
1949, and continued for approximately one month. On the basis of the
survey, Six sites were selected for excavation. Excavations were commenced
in January, 1950, and were concluded in the early part of March. During
the following month, the pottery and artifacts were processed and then
attention was directed to Tinian.

On Tinian, the necessary survey was greatly facilitated by the fact that
the United States Geological Survey had a three-man team working on the
island; these men had in the course of their own work discovered virtually
all the major sites. Their co-operation and aid were invaluable and greatly
expedited the work. Following the Tinian survey, two sites were selected
for excavation. The Tinian work was completed in May, and the pottery
and artifacts were shipped to Saipan for processing.

In June, through the co-operation of the United States Navy, transpor-
tation was arranged to Rota and a brief archaeological survey was made
of that island.

Neither the expedition’s budget nor the amount of time available al-
lowed large scale excavations. On Saipan, a four-man digging crew was
employed and on Tinian a two- to three-man crew. Every man proved to be
excellent and set a pace of work that was remarkable in any climatic zone,

INTRODUCTION 27

let alone the tropics. In no case, however, was a single site completely exca-
vated. Instead, an attempt was made to obtain the principal structural
characteristics of each site and to seize any opportunity that gave promise
of stratigraphy.

A word should be said regarding excavation methods. Each site was first
cleared of bush. It was then staked on a grid with a five-foot grid interval,
except for Laulau Rock Shelter on Saipan, where a system described later
was employed. Each site was excavated in 0.5 foot levels. Maps were con-
structed with plane table and alidade.

Pottery, artifacts, and burials were cleaned at the expedition quarters by
a washing crew. Except for analysis and counting of Marianas Plain sherds,
some 7,000 of which were counted and discarded, all material was packed
and shipped to the Museum for final study.

II. Survey and Excavations on Saipan

PHYSICAL CHARACTERISTICS OF SAIPAN

Saipan is approximately 12.5 miles long and 5.5 miles across at its widest
point; the land area is a little over 46 square miles. The island consists
primarily of a series of raised coral limestone terraces on a volcanic base.
Saipan can be divided into two major surface areas: a mountainous interior
upland, which occupies nearly a quarter of the total land area; and a series
of plateaus and coastal terraces, and a low coastal strip on the western
side, which surround the rugged interior. The more level coastal areas pro-
vide the best farmland. It is in these areas that most of the archaeological
sites lie. Particularly important is the western shore area with its adjacent
lagoon, for the fish of the lagoon have in historic times—probably in pre-
historic as well—been a principal food of the islanders. In historic times
the main centers of population have been on the lagoon side of the island.

DESTRUCTION OF ARCHAEOLOGICAL SITES

Several factors have contributed to a destruction of archaeological sites
on Saipan. During the last decade of the Japanese regime, the agricultural
resources of the island were developed to the utmost for the growing of
sugar cane. Probably most of the arable land was cleared and planted.
Although the Japanese did not introduce large amounts of power machin-
ery in their farming operations, and though plowing was generally shallow,
considerable disturbance of sites did result. In addition, the village of
Garapan, located on the western shore, expanded to a sizeable town of
some 13,000 people. A large sugar mill and an accompanying village were
erected at nearby Chalan Kanoa, and the western shore area was generally
utilized in the expanding economy of the island, so that such sites as from
hearsay evidence existed were largely destroyed.

The pre-World War II disturbance of archaeological sites was of much
less significance, however, than the destruction occasioned by the invasion
of Saipan during World War II and the construction of base facilities fol-
lowing the American occupation. According to local informants, as well as
the fragmentary literature, most of the principal archaeological sites were
located on the western and southern coastal areas. The western area in par-
ticular was the scene of bitter fighting during the invasion. Even more im-

SURVEY AND EXCAVATIONS ON SAIPAN 29

portant, however, was base construction. Great areas were bulldozed and
then covered with crushed limestone to provide foundations for ware-
houses, troop quarters and airstrips. The entire coastal terrace from Agin-
gan to Cape Obian was transformed into a giant ammunition dump, with
virtually all the top soil bulldozed into revetments. As a result of the
enormous amount of wartime construction, probably most of the archaeo-
logical sites on the island have been destroyed. The Museum’s archaeologi-
cal work was, therefore, considerably a salvage operation. However,
enough sites were found to make excavations productive.

SURVEY

Archaeological survey was commenced at Agingan Point. From this lo-
cation, the beach, cove, and coastal terrace areas were surveyed in com-
pany with local Chamorro residents. The survey proceeded in counter-
clockwise fashion around the island. All beach and cove areas were cov-
ered. Even with the destruction of sites, it soon became apparent that the
major archaeological sites were located adjacent to beaches and coves in
spots where ancient populations had had access both to the resources of
the sea and to fertile farmland. Most of these locations were used as mili-
tary areas during World War II and have been bulldozed, but numerous
potsherds scattered on the surface, and /atte stones removed to the sides of
roads and former military areas remain as testimony of the previous exist-
ence of sites. No major /atte site exists completely intact. The site at Agin-
gan described by Hornbostel (Thompson, 1932, p. 19) has been destroyed.

Except along the shore line, in cultivated areas, and in those covered by
groves of casuarina, survey proved to be very difficult. Dense secondary
growth of bush, heavy grass, and abandoned sugar cane made inspection
of the ground surface a formidable task.

Some twenty-five limestone caves were also investigated. Only one small
one was found undisturbed. The others had been excavated by Japanese
defense forces and converted into machine gun positions and military
strong points for the defense of the island during the American invasion.
Saipan, however, abounds in limestone caves and it is probable that a
number still exist undisturbed.

Principal sites located during the survey are as follows (see fig. 5):

1. Objan: The coastal area between Agingan Point and the base of
Naftan Point contains numerous sherd areas and bulldozed remains of
latte house sites. Only one was found intact, fortunately spared destruction
by its location between two large outcrops. This atte is at Objan and is part
of a large site that has been bulldozed. The Objan /atte was excavated and
further details are given in the description of the excavations.

SABANETA PT.

145° 50'

g
MAIGO
i a alk ee a FAHANG
FATANAPAG
MUGHOT
PT.
\5
LAULAU
L— 15*10! KATAN 15°10" —
PT.
30" -
susUPE 9 BS)
CHALAN|E S 1ise°9' —
KANOAT HS HANGMAN PT.
F Ly HH
AB HIASLITO MAGICIENNE. BAY
( sm)
DANDAN PT.
AGINGA
PT.
CAPE OBIAN z
NAFTAN PT.
145° 50'
1 (eae ay.

Principal archaeological sites of Saipan. Numbers on map refer to fol-

lowing sites: 1, Objan; 2, Laulau; 3, Bapot; 4, As Teo; 5, Talofofo; 6, Fafiunchulujan;

7a, Oleai; 7b, Chalan Kija; 7c, Chalan Piao; 8, Chalan Galeite.

SURVEY AND EXCAVATIONS ON SAIPAN 31

BAPOT SITE

LAULAU
SITIE

Bapot Beach

Laulau Beach

MAGICIENNE BAY
(LAULAU BAY)

| Mile (Nautical)

Fic. 6. Magicienne Bay, Saipan, showing location of Laulau and Bapot Sites.

2. Laulau (see fig. 6): There is an extensive village area back of the
beach on Magicienne Bay. This site was undisturbed by military opera-
tions and was partially excavated. Further details are included in the sec-
tion on excavations.

3. Bapot (see fig. 6): North and east along the shores of Magicienne Bay
there is a coastal terrace about 100 yards wide. The soil is good and there is
easy access to reef and offshore fishing. This area is the site of three clusters
of latte houses: Bapot I consists of two, Bapot II of four, and Bapot III of
five. All have been disturbed by defensive trenching by the Japanese mili-
tary forces, while a road also cuts across the former occupation area. Sherds
are distributed throughout the area. It is probable that the area once con-
tained numerous house sites.

4. As Teo: An inland site, located at the base of a cliff, where two springs
emerge and develop as very small streams. The site consists of a large sherd
area about four acres in extent, but with no surface indications of refuse
dumps or house sites. As Teo is typical of sherd area sites found near water
supplies along inland parts of the eastern section of the island.

5. Talofofo: At Talofofo a small stream enters a coastal cove through a
steep-sided canyon. On the north rim of the canyon there is a sherd area,

32 MARIANAS PREHISTORY

Road on old Cliffs
Jopaonese RR

right of way TSUKIMI 1.

(Isleta Maigo Fohang)
Cliffs FANUNCHULUJAN BAY

|
! Mile (Noutical)

Fic. 7. Fafiunchulujan Bay, Saipan, showing location of Fafunchulujan Site.

but most of the site has been destroyed by the construction on it of wartime
troop quarters.

6. Fanunchulujan (see fig. 7): The site is located along the top edge of
the cliff-like sides of Fanunchulujan Bay. The site consists of a string-like
sherd area approximately 150 yards long, together with a single latte house.
Fragments of broken /atte stones are also found at the edge of the site,
which had been plowed for sugar cane during Japanese times. Evidently a
small village had been built along the edge of the cliffs.

7. Western coastal area along the lagoon shore: According to Chamorro
testimony, this region once abounded in sites. However, its development as
a major occupation area during Japanese times and as a wartime construc-
tion area has resulted in the destruction of most of these sites. Today, sherd
areas are to be found intermittently along the entire western coastal strip in
locations of lesser disturbance. Tanapag village itself was an old site, judg-
ing from the sherds to be found on the small plaza. Oleai, Chalan Kija, and
Chalan Piao are other sites in this area, each of which was excavated and of
which details may be found in the section on excavations.

4
= ee - “
Zo hey,
eed | = - see.
T AS brie a
MAG SF f hen
g
N ck igen
“ye sleet
Oe, ya
Se4 Fe -
Aa Ses
y- ge =
ZA
wo

Limestone

Outcrop

\
ae —7/l | Fn
aes =
+ 2 oe
r ES =
ieee =
+ ig se Gee aS
has ava dee re Ne
ae ee es
fd dates g Nees Sa A ee
ean eee =
| ee -
La ae sLo¢ (Us eae eS ©
: y 4 intrusive oie : eA ¥
| Li Pits = eee a
oD Oe See. SN —
: Sea? ,
co

| 8 | °90 ie \
¥—--f-

pyre

ot ~
rae eae Braces \ srk
BB + 7 \ \ ; 2
j : me on
7 5! la I \ ' a
’ I Fa ' \ ~ =
3 : a
o yes 5 eects th ' . & -
25° 80 18 Or SO =..._| A aha
gd eee eee tee 3 a
ares v =z
Sais re
Sos ° ae
See ra 2
Limestone SS Se eee
~ Es ‘= ~
Ouvtcrop Seca Ss 2
Sees joe
eran =
Ore SS 10 Pree =
—-.-:.290 =e —
[ HHH J aa ce
FEET een ne ae
Fic. 8.

Plan of atte and excavations at Objan, Saipan. Broken lines indicate limits
of excavations; shafts are numbered; capstones are lettered.

oe)

34 MARIANAS PREHISTORY

8. Chalan Galeite: Although upland sites in the form of small sherd
areas exist on the island, the number of people residing in interior settle-
ments must have been small. Chalan Galeite is the only upland, interior
latte site found. It consists of two small latte houses and a small sherd area.

As a result of the survey, only six, partially intact, latte sites were found:
Objan, Laulau, Bapot, Fanunchulujan, Chalan Kija, and Chalan Galeite.
Three of these were selected for excavation: Objan, Laulau, and Chalan
Kija. Sherd areas are much more numerous than /atte sites, but in no case
in sherd areas was there surface evidence of undisturbed dump heaps and
remains of houses made of perishable posts and materials.

EXCAVATIONS AT OBJAN
LOCATION AND DESCRIPTION OF THE SITE

Objan, also known locally as Obian, is a large village site located about
midway between Cape Obian and Naftan Point. The site lies directly be-
hind a long beach (Unai Objan or Unai Obian). At this point there are no
steep cliffs fronting the beach, and the low terrace on which the site is lo-
cated slopes toward the beach. The beach in turn leads onto a fringing
reef. Inland from the site, the soil is shallow but good. The location of the
site provided, for its former inhabitants, easy access both to soil resources
and to reef and offshore fishing.

The habitation area extends continuously for approximately 1,200 feet
along the shore. The original width of this area is difficult to determine, for
as a result of wartime bulldozing, the top levels of the site have been
scraped off and dumped seaward. The site was probably not more than 300
feet wide. Today the surface is covered with sherds, shells, and, in places,
fragments of latte stones.

At about the midpoint of the site, directly on the beach, the Japanese
military built a concrete pill box. In addition, the Japanese constructed a
narrow gauge railroad across the beach side of the site. The railroad right
of way (ties and rails have been removed) cuts across part of the site at
both east and west ends.

Objan has therefore been much disturbed. However, one section of the
site was found intact. This consisted of a single /atte house, located 225 feet
east of the Japanese pill box, and situated on a low sandy terrace between
two large limestone outcrops (figs. 8, 9). The railroad right of way cuts
directly in front of the house and has destroyed its southeast corner. The
beach extends to the right of way; it is approximately 150 feet from the site
to the mean high water line.

At the time of excavation, the site was covered with heavy grass and a
few small trees. North of the /atte, the two limestone outcrops join and at

SURVEY AND EXCAVATIONS ON SAIPAN 35

this point there is an abandoned Japanese machine-gun post, though the
latte do not appear to have suffered wartime damage. The house site com-
mands a fine view of the sea and of Tinian Island, some four miles distant.
The limestone outcrops serve as partial protection from heavy winds, with-
out completely cutting off the cooling trade breeze.

Objan is of particular interest because it is identified in seventeenth cen-
tury Spanish accounts. In January, 1670, Medina, a Spanish Jesuit mis-
sionary, crossed over the passage between Tinian and Saipan, landed at
Objan, and proceeded inland. While attempting to baptize a child at one
of the interior villages, he was attacked and killed (Le Gobien, 1700).

In 1684, a Spanish expedition, led by José de Quiroga, landed on Sai-
pan’s west coast and after considerable fighting marched south to the vil-
lage of Agingan, on the south shore. This village undoubtedly is the large
site described by Hornbostel (Thompson, 1932, p. 19), though it has since
been destroyed. The Spanish party then sent friendly Chamorros to Objan.
It is noted in the Spanish account that the people of Objan had previously
been cordial to the Spanish.

Shortly after, there was a general revolt of Chamorros in the Marianas
and Quiroga found himself besieged on Saipan. He counterattacked with
his small party, sacked Objan, and procured eight canoes from the village
to take him to Guam (Le Gobien, 1700). In 1698, the Spanish sent another
expedition to Saipan, were finally victorious, and forced the Chamorros to
move from Saipan to Guam. It is highly probable that Objan was aban-
doned at this time.

EXCAVATION PROCEDURE

Although the construction of the railroad right of way along the south
side of the house had destroyed a considerable area of cultural deposit, the
right of way served as a convenient starting point for obtaining a profile
of the soil structure underlying the house. The first step in excavating
the site was the cutting of a profile along the bank formed by the right of
way. This profile is indicated by the broken line XX’ shown on figure 8.
The small area of deposit removed by the cutting of the profile was not
excavated in levels, as most of the cultural material had been secondarily
deposited down the slope of the bank.

Following the cutting of the profile XX’, an east-west trench, 5 feet wide
and 22 feet long, was excavated at the west edge of the house, and a second
trench, of the same width and 16 feet long, was excavated along the north
side of the house. Both trenches were carried down either to bedrock or
into undisturbed sand. Excavations were then carried forward into the
house proper (figs. 8, 10). An important consideration was that an agree-
ment was made with the Navy Civil Administration not to destroy the site

36 MARIANAS PREHISTORY

as a result of the excavations, for the Administration desired that the latte
be preserved, in view of the fact that so few latte sites on Saipan remain
intact.

STRUCTURE OF THE SITE

General Description

The principal feature of the site was a series of nine stone shafts and as-
sociated capstones. So far as could be determined, there were originally ten
shafts, the one at the southeast corner having been destroyed by the con-
struction of the railroad right of way. Only five shafts were found erect and
in place; the remaining four had weathered away at ground level, so that
the bases remained in place but the exposed part of each shaft had fallen.
All the capstones had long been displaced and most of them were broken.

As measured along the shafts, the house was 46 feet long and 12 feet
wide, with the shafts placed from 7 to 9 feet apart. Sherds were found
scattered on the surface around the latte.

The subsurface structure of the site proved to be relatively simple. The
top stratum was composed of gray sand, varying within the /atte area from
0.5 to 1.0 foot in thickness. This stratum was probably, to a large extent,
wind-blown. On windy days, sand from the beach blew into our faces. In
addition, sand has undoubtedly been washed down from the area above the
site. Below the top stratum lay a thin layer, approximately 0.2 to 0.5 foot
thick, of coral pebbles. This layer was present within the area enclosed by
the stone shafts and extended toward the seaward side of the house. The
coral pebble stratum was visible on the XX’ profile (see fig. 12). I would
interpret this coral pebble layer as representing the ground surface at the
time the house was occupied. Today in the Marshalls, as well as in many
other parts of Micronesia, it is customary to spread coral pebbles under and
around houses. The coral pebble surface prevents mud from forming in
rainy weather and reduces dust during dry spells. In the case of Marshall-
ese houses raised on posts above the ground, the area under the floor be-
comes a work area, particularly for the women. This area always has a
coral pebble floor. The /atte house at Objan apparently possessed this com-
mon Micronesian feature.

Below the layer of coral pebbles was a thick stratum of very dark gray
sandy soil, filled with much organic material, including many minute bits
of charcoal and containing numerous potsherds, artifacts, limestone and
coral rocks, and burials. Between the northern line of shafts and the XX’
profile, this stratum was found to vary from 1.5 to 5.5 feet in thickness.

The very dark soil stratum was underlaid either with limestone bedrock,
or with a stratum of gray sand, containing some sherds. This stratum in
turn merged into underlying white beach sand, sterile of artifacts.

to excavation.

ion, prior

top of shaft has weathered away.

Objan J/atte cleared of surface vegetat

9° Upper:
: shaft 5; looking north

Fic.
Lower

38 MARIANAS PREHISTORY

Fic. 10. View from above of excavations at Objan; looking southwest.

No evidence was found of any house structure under the /atte. Yet the
depth of deposit underneath the coral pebble layer indicates a considerable
occupation period prior to that of the Jatte house. In addition, the base of
shaft 9 was found resting on a deposit of ashes and sherds 1.1 feet thick,
another bit of evidence indicating prior occupation.

During the course of excavation, eleven burials were exposed. These
were found either within the shaft area, or directly to the west and south
of the house. The burials lay between 1.1 and 4.0 feet below the present
ground surface. Although only two burials showed evidence of the pit in
which the bones had been deposited, it is probable that all burials are
referable to the latte house occupation.

Details on the shafts and capstones, the soil structure of the site, and
burials follow.

SURVEY AND EXCAVATIONS ON SAIPAN 39

Fic. 11. Shaft 7, Objan, Saipan; looking west. Shaft has fallen and has broken off
at base; base remains in place.

LATTE

Shafts

Weathering. —All latte shafts exhibited pronounced weathering—those
that had fallen more so than those that remained erect. Shafts 1-5 were in
place and erect, though no. 4 was leaning precariously. Shafts 6—9 had
fallen. The bases of these shafts remained in place, but the shafts had
eroded away at the ground surface and the upper sections had fallen to the
ground (figs. 9, 11).

Material.—Coral limestone.

Form and Dimensions.—All shafts were clearly trapezoidal in shape. Shafts
2 and 3 exhibited the least amount of weathering and most clearly illus-
trated the original form of the shafts at the site. Shaft 2 was 3.0 feet wide
and 1.7 feet thick at the present ground surface, and 2.1 feet wide and 1.6
feet thick at the top of the shaft. Shaft 3 was 2.9 feet wide and 1.7 feet thick
at the ground surface, and 2.2 feet wide and 1.5 feet thick at the top. The
overall length of shaft 2 was 6.7 feet, while that of shaft 3 was 7.0 feet. So
far as could be determined, the shafts at the site originally averaged be-
tween 6.0 and 7.0 feet in length. The ground surface at the time of occupa-
tion of the house was approximately 1.0 foot below the present surface at
the shaft locations. At this time, shaft 2 projected 4.7 feet above ground
level and shaft 3 projected 5.2 feet. The tops of the shafts were approxi-
mately level with each other.

Constructional Features —The depth of the shafts below the present ground
surface varied from 2.0 to 3.0 feet. The coral floor that probably repre-
sented the original ground surface lay approximately one foot below the

40 MARIANAS PREHISTORY

present surface, so the shaft must have been set in holes from one to two
feet deep. Shaft 6 was set in the shallowest hole, the base resting on lime-
stone bedrock. The variation in depth of the shafts was compensated for by
their variable length and by the variable thickness of the capstones. The
base of each shaft was packed with large rocks, which helped to keep it
erect. The base of shaft 9 was resting on an ash deposit 1.1 feet thick, below
which was coral limestone bedrock.

Capstones

Weathering —Only four capstones, recognizable as such, were dis-
covered. Numerous rocks were found on and around the site and pre-
sumably some of these represented the disintegrated remains of the other
capstones. The four capstones exhibited pronounced weathering. Capstone
D, the best preserved, was found buried in the sand on the seaward side of
the right of way.

Material.—All capstones of coral limestone.

Form and Dimensions.—Al]| capstones conformed to the general Marianas
shape. Capstone D, the least affected by weathering, was 2.5 feet thick and
from 4.1 to 4.2 feet in diameter. The remaining three capstones varied
from 2.3 to 2.7 feet in thickness, and from 3.8 to 4.2 feet in diameter.

Sot PROFILES

Five east-west and two north-south profiles were cut. The accompanying
two profiles illustrate the soil structure of the site (fig. 12). The upper
stratum probably represents wind-blown and washed-in sand that was de-
posited after the house was abandoned. The next lower stratum is a coral
pebble layer, which, judging from its occurrence in other parts of Microne-
sia at the present day, was a surface laid down by the house’s occupants as a
gravelled work area. This coral pebble layer is confined to the space within
the shafts and to the seaward side. Below the coral pebble layer is a stratum
of dark gray sand, filled with much organic material, rocks, potsherds,
artifacts, and burials. Below this stratum is either coral limestone bedrock
or a gray sand stratum. The limestone bedrock has a very uneven surface,
pitted with many natural holes and depressions. The gray sand stratum is
sharply marked off from the very dark stratum overlying it by its color, lack
of small rocks, and the relative absence of charcoal bits and finely ground
organic matter. The gray sand stratum contains some sherds. This stratum
merges into the underlying white coral beach sand, which contains no cul-
tural material.

BurIALS

During the course of excavations, eleven burials were exposed. These

were located either under the house, on the seaward side of the house, or

SECTION XxX!

"9"@, G05 ,0, 0
SOON

SNOW

xO
rap aA

ea’
1 As: oy ova en w7, Y¥evav
eS? - oA <5 re tA vav7y
Laas eae PATA BC eA Ia CPR Fb ag had Vag k TV > cy a oA
A ee ge bie VIET TS ae Aie ype Ae ep peek et agg Vv OME ss
PUR CN 2 C7 Velo Nia ket eA egy b

SEGTLOIN, 22Y X77

Light gray sand

Coral pebbles

Dark gray sand

Al Light gray sand
ey White beach sand, sterile of artifacts
Fic. 12. Soil sections, Objan, Saipan.

42 MARIANAS PREHISTORY

directly to the west of the house. The number of burials was large, consider-
ing the limited area actually excavated, and it was clear that the under-
ground structure of the house site was literally filled with burials. Limita-
tions of time did not permit the complete excavation of all burials ex-
posed, but it was possible to determine that a variety of burial practices
existed.

All skeletal material was in fragmentary and poor condition. In three
cases the skulls were not present—evidence of the removal of the skull after
death and before inhumation, a custom documented in the early accounts.
In several instances, the teeth were stained dark red-brown, an indication
of the antiquity of betel-nut chewing. No grave goods were found with any
of the burials, with the single exception of the probably accidental associa-
tion of one net-sinker with Burial 6. The coral pebble layer, which I have
interpreted as the floor of a work area, was extended, unbroken, over the
burials beneath it.

The eleven burials were of the following types: two secondary burials;
one probably extended on back, shoulders to north; one extended on side,
shoulders to north, facing east; one extended on side, shoulders to south,
facing west; one on back, with legs drawn up, shoulders to east; five in-
determinate.

Burial 1: Located seaward side XX’ profile. Secondary burial, with mis-
cellaneous, fragmentary collection of bones lying in bottom of pit. Top of
burial 3.2 feet below present surface.

Burial 2: Located seaward side XX’ profile. Child burial lying in pit.
Sudden cave-in of bank prevented detailed determination of type of burial.
Top of burial 2.1 feet below present surface.

Burial 3: Located seaward side XX’ profile. When the profile was cut,
the skull of this burial was exposed on the face of the profile. Surface depth
to top of skull 3.3 feet. No attempt was made to excavate this burial.

Burial 4: Located seaward side XX’ profile. When the profile was cut,
two long bones were partially exposed on the face of the profile. Burial lay
2.9 feet below present ground surface. No attempt was made to excavate
this burial.

Burial 5: Located in square 15-15AA. Skeleton on back with legs drawn
up; femurs extended; feet at pelvis. No skull present. However, a mandible
was found at the position of the clavicles. The mandible may represent a
second individual and could have been interred as a fetish. Top of burial
4.0 feet below present ground surface.

Burial 6: Located in square 15DD. A shallow secondary burial consist-
ing of a miscellaneous assortment of bones, including three fragmentary
skulls. Burial lay just under coral pebble floor. Top of burial 1.1 feet below
present ground surface.

SURVEY AND EXCAVATIONS ON SAIPAN 43

Burial 7: Located in squares 10BB-15BB. Only a skull, lying on its right
side, was exposed by the trench excavated along the 15-foot stakes. No
attempt was made to excavate the entire burial. Top of burial 3.0 feet
below present ground surface.

Burial 8: Located in squares 15BB-15CC. Only bones of two feet, ex-
tending into trench, were exposed. Remainder of burial was not excavated.
Burial probably extended on back, with skull to south. Top of burial 3.3
feet below ground surface.

Burial 9: Located in squares 10DD-15DD. Only the upper part of the
skeleton was exposed, projecting into the excavated area. The lower ex-
tremities were not excavated. Burial lay on back, with shoulders to north.
Burial was probably extended on back. No skull present, but a mandible
was found at the clavicles. This burial lies directly below Burial 6. Top of
burial 2.2 feet below present ground surface.

Burial 10: Located in squares 5-10. Burial extended on left side, shoul-
ders to north, facing east. No skull present. Top of burial 2.6 feet below
present ground surface.

Burial 11: Located in squares 10DD-15DD. Burial extended on left side,
shoulders to south, facing west. Lower extremities lay on limestone bed-
rock. Burial 2.6 feet below present ground surface.

EXCAVATIONS AT LAULAU
LOCATION AND DESCRIPTION OF THE SITE

Like Objan, the Laulau Site is a coastal one, permitting the combination
of good farmland with access to the fish resources of the sea (see fig. 13).
The Laulau Site is located on the northwestern shore of Magicienne Bay,
on the east coast of Saipan. The site lies about 150 feet inland from the
beach, to the west of a narrow road that runs around Magicienne Bay.
Laulau Site is approximately 600 feet long and 200 feet wide, with its long
dimension paralleling the shore line. It is northwest of a massive, concrete
Japanese blockhouse built on the beach and commanding the bay.

Northwest of the site there is a series of limestone cliffs. A large limestone
segment has fallen away from the cliff edge and forms a rock shelter. This
is Laulau Rock Shelter, whose excavation will be described shortly. The
ground at the rock shelter is approximately 115 feet above sea level. From
the rock shelter the terrain slopes gradually down to the beach, and the site
is located on this sloping ground.

The surface features of the site consist of a large number of sherds scat-
tered over the surface, plus the remains of four /atte houses, all badly dis-
integrated. Two Jatte houses are located at each end of the site. The four
houses are in rough alignment. It is quite possible that other atte houses

MAG
N .. HOUSE
- D
-+ HOUSE
a5 C
~
Sherd Area ae
ah
S
=
ie
— “ee
oR f
% LEEDS
iy, Wz
Bare ae Japanese
eae Ge o Blockhouse
== Se o,
= oh Sherd Area es
ROCK SHELTER
s
\
\
~ ay ale
x N
aos - 4+ HOUSE PACIFIC
oe ee: OCEAN
a i ay
oy arg
= oe Beach
Sherd Area z " HOUSE
Cx nn
50 0 50 100
SeSee: amen
REET
Fic. 13.

Plan of Laulau Site, Saipan.

SURVEY AND EXCAVATIONS ON SAIPAN 45

once existed, but have since disintegrated (see fig. 13). At the time of exca-
vation some of the site was planted in maize, coconut palms, and bananas,
and the rest was lying fallow in grass. According to Chamorro informants,
the land has always been hand-tilled. In Japanese times it was used for
tobacco. Today, it is used primarily for growing maize.

The beach at the point of the site is bordered by a wide, fringing reef. As
the location is on the trade wind side of the island, the waters of Magi-
cienne Bay are usually rough. In early times it would have been difficult
to bring canoes of any size over the fringing reef; however, the reef fishing
is excellent.

At the Laulau Site, two structures were excavated: House A, the south-
westernmost /atte house; and the rock shelter. Description of their excava-
tion follows.

House A

Excavation Procedure

The area immediately surrounding the site had been cultivated but was
lying fallow at the time of excavation. The site was cleared of grass and
staked according to a 5-foot grid (see figs. 14, 15). A longitudinal trench,
2.0 feet wide, was then run through the long axis of the site. Transverse
trenches were excavated at each end of the site and were widened to obtain
necessary structural details. All excavations were carried down into undis-
turbed soil. The limits of excavation of the house proper are shown in
figure 14.

A common contemporary Micronesian house pattern consists of a living
house, with a separate cook house to the rear of the living house. In order
to determine whether House A possessed this feature, an area to the north-
west of the house proper was excavated. This latter area is not shown on
figure 14. It comprised squares 15D, 15E, 15F, 15G, 20E, 20F, 20G, and
20H, whose location can be judged by an inspection of figure 14.

Structure of the Site

House A consisted of ten shafts arranged in two parallel rows of five
shafts each. The shafts were very badly weathered, though the shaft bases
remained in place. On the surface of the site was a high concentration of
scattered sherds, mixed, along the seaward margin of the house, with large
numbers of coral pebbles. At a depth of 0.3-0.5 foot below the surface,
along the seaward side of the house, areas of a coral pebble stratum were
discovered. This coral pebble stratum was approximately 0.2 foot thick.
It extended into the house for 1.5 feet and outward for approximately 5.0
feet. This was no doubt the remains of a coral pebble working area, similar
to the one at Objan. However, at House A the coral pebble stratum was

46 MARIANAS PREHISTORY

confined to the seaward margin of the house, while modern tilling of the
site had brought many of the floor pebbles up to the present surface.

Another indication of the ground level at the time the house was occu-
pied was the discovery of a large section of shaft 4 that had broken from the
shaft and crushed most of a pot beneath it. The sherds were lying on a sur-
face 0.5 foot below the present ground level. This lower surface probably
was approximately the ground level at the time of house occupation,
though this level was nowhere found well defined.

In the area excavated northwest of the house proper, remains of two
earth ovens were found, though there was no evidence of a structure that
would have served as a cook house. One oven was discovered in square
15F, the other in square 20E. The former consisted of a circular shallow de-
pression 2.4 feet in diameter, filled with small, heavily burned rocks. The
top of the deposit lay 0.7 foot below the present ground level. The second
oven was similar—2.2 feet in diameter and 0.8 foot below present ground
level. The Chamorro and Carolinean digging crew recognized these im-
mediately as earth ovens, which are still used occasionally on Saipan today.

Also in this area was found a stone mortar (fig. 72). It was 1.4 feet thick
and projected 0.5 foot above present ground level, but its base was also
buried 0.5 foot below the probable former ground level. Another mortar
was found at the south edge of the house, but it had obviously been second-
arily deposited in the position in which it was found.

Another discovery of interest was two iron fragments, at the northeast
corner of shaft 9, lying next to the shaft and directly under a large stone.
This large stone was one of several that had been packed around the base
of the shaft to help keep it erect. The iron fragments were found at a depth
of 1.0 foot. None of the shaft bases had been disturbed by modern tilling
and there was no evidence that the iron fragments were intrusive. One
fragment was a section of an iron knife; the second was a projectile point
(fig. 89).

Unlike Objan, no burials whatsoever were found associated with the
house, a very unusual feature for atte houses in the Marianas. The shallow
depth of the occupation level at the house indicates that it was inhabited
for only a relatively short time.

P The Latte
HAFTS

Weathering.—All shafts exhibited extreme weathering. Most of them
were leaning or had fallen, though shaft bases remained in place.

Material.—All shafts of coral limestone.

Form and Dimensions.—Weathering of shafts was so great that the only
real indication of form was found in the buried shaft bases, which were
clearly rectangular. Those shafts which were approximately whole varied

MAG
yy ey

FEET

Fic. 14. Plan of latte and excavations, House A, Laulau Site, Saipan. Shafts are
numbered; capstones are lettered. Limits of excavation indicated by broken lines.

Fic. 15. House A, Laulau, Saipan, looking northeast. Upper: site cleared and
staked. Lower: site under excavation. Corn field and coconut palms in background.

Fic. 16. House A, Laulau, Saipan. Upper: shaft 10 in foreground; shaft 1 in
background; capstone fragment lies next to shaft 10. Lower: shaft 4. Rocks packed
around bases of all shafts.

50 MARIANAS PREHISTORY

in dimensions as follows: overall length, 3.1-3.6 feet; width, 2.2—3.0 feet;
thickness, 1.2—2.4 feet.

Constructional Features—So far as could be determined, the ground sur-
face at the time of the occupation of the house was from 0.3 to 0.5 foot below
the present surface. The /atte shafts were set in shallow holes from 0.3 to
0.6 foot below the old ground level, and the shaft bases were packed with
rocks to assist in maintaining the shafts erect (see fig. 16).

SECTION XX’

10 feet

Dark gray top soil
Py Red brown subsoil, sterile of artifacts

Fic. 17. Soil section, House A, Laulau Site, Saipan.

CAPSTONES

Weathering.—Six capstones, or recognizable fragments thereof, were
found near the shafts from which they had been displaced. No capstones
were found with shafts 3 and 4. Apparently these capstones had disinte-
grated completely. All capstones exhibited pronounced weathering.

Material.—All capstones of coral limestone.

Form and Dimensions.—Capstones E-H were the best preserved and ex-
hibited the usual Marianas cup-shaped form. These capstones varied in
diameter from 2.4 to 2.8 feet, in thickness from 1.4 to 1.8 feet.

Soil Profiles

The soil profiles at House A, as exemplified by that shown in the illus-
tration (fig. 17), are very simple. The top stratum, from 0.3 to 0.8 foot in
depth, consists of dark gray, humus-filled soil, containing shell fragments,
potsherds, and all other cultural material found at the site. The lower
stratum, of undetermined depth, is a reddish-brown, clay-like subsoil,

Limestone

------- — PRIMARY OVERHANG

Ty oe SECONDARY OVERHANG
=] TRENCH
—=i———* EXCAVATION

Fic. 18. Plan of excavations, Laulau Rock Shelter, Saipan.

52 MARIANAS PREHISTORY

completely sterile of artifacts. Along the seaward side of the house, the top
stratum contained numerous coral pebbles, and fragments of a pebble floor
were found in this area, at a depth of 0.3-0.5 foot below the present ground
surface. This floor, lying at such a shallow depth, undoubtedly lost its well-
defined character through modern tilling of the land.

LAuLAU Rock SHELTER
Excavation Procedure

This rock shelter was formed by the overhang of a large limestone section
projecting from the seaward bluff of a limestone terrace, now many feet
above sea level. In the northern part of this limestone fragment was a cave
which had been completely cleaned out by the Japanese military forces and
used as a military strong point. The part of the site excavated by the Mu-
seum lay along the southern side of the limestone under the overhang of a
cliff face (see fig. 18). This area had not been disturbed.

A somewhat unorthodox method was used in laying out the excavation
grid. The main axis was run from a point outside the shelter into its deepest
part. Lateral grid lines were then staked parallel to the rock face of the
shelter, rather than at right angles to the main axis. This resulted in more
or less trapezoidal horizontal sections. Excavations were commenced be-
tween the 20- and 24-foot stakes, and between the A and DD stakes. This
trench, paralleling in long dimension the rock face, was then carried for-
ward toward the face (fig. 18). The unorthodox grid made for convenience
in excavation and corresponded to the occupation area, which was spread
around the curving face of the cliff. The site was excavated in 0.5 foot
levels, which paralleled the ground surface.

Structure of the Site

Laulau Rock Shelter was primarily a burial site, and the structure of the
site revolved around the relative position and nature of the burials exca-
vated. The principal features of the site were as follows:

(1) The surface was devoid of any indication of house construction, and
except for a few sherds there was no evidence of human occupation. The
top stratum consisted of a thin layer of humus-filled, dark gray topsoil,
approximately 0.1—0.3 foot thick.

(2) Below the thin stratum of topsoil lay a stratum of densely packed
ashes, approximately 1.5 feet thick. This ash stratum was filled with frag-
ments of charred human bones. Three burials—one extended, one flexed,
and one a secondary burial—lay in shallow pits in the ash level, but other-
wise there were no discernible structures within the ash level itself. The ash
bed extended from 10 to 18 feet out from the rock face of the shelter; the
outer margin of the ash bed curved roughly parallel to the rock face.

Fic. 19. Laulau Rock Shelter, Saipan. Upper: excavations in progress; looking
north. Lower: excavating along the rock face.

54 MARIANAS PREHISTORY

The ash bed contained so many small fragments of bones that the only
conclusion that can be reached is that this stratum represented the remains
of a large number of human cremations. A certain amount of pottery—all
sherds—was found in the ash level. These sherds could have come from
pottery placed with the dead at the time of cremation or otherwise associ-
ated with disposal of the dead.

(3) A light brown sandy surface lay directly beneath the ash bed. This
surface had very obviously been spread with beach sand. At this level three
scattered postholes were found, but no other evidence of a house structure;
the three postholes were not in a regular pattern and a wall could not be
delineated from their location.

The sandy surface merged into the brownish soil beneath, making a
single stratum. This stratum was difficult to follow. At times it was only 0.3
foot thick; at other times it increased to several feet in depth.The reason
for this vertical irregularity was that the stratum merged with the fill of a
considerable number of large burial pits that had been sunk into the
ground. Most of these burial pits contained extended burials, only one of
which showed evidence of cremation.

(4) The large burial pits had been excavated into a heavy clay subsoil,
sterile of all artifacts.

The structure of the site is shown graphically in the accompanying soil
profile (fig. 20). It is clear that two levels of human use are involved. The
upper level is represented by the ash bed, the remains of an unknown num-
ber of cremations. The lower level is represented by a sandy surface associ-
ated with underlying burial pits, exhibiting a different burial practice.
Furthermore, below the sandy surface, Marianas Red and lime-filled, im-
pressed sherds were found; none of these were present in the ash bed.
Further consideration of the pottery differences found at this site are in-
cluded in a later section. Details regarding the burials follow.

Burials

The upper stratum of ash contained such a profusion of small charcoal
fragments of human bones, scattered in random fashion through the stra-
tum, that it was not possible to separate these into well-defined cremations.
Sections made through the ash failed to reveal any lenses within the
stratum. Whether this means that the cremations responsible for this stra-
tum all took place in a relatively short time, is difficult to say. Just how
many individuals are represented could not be determined.

Three burials were found in the ash level. They were all intrusive into
this level, though how much Jater they are than the cremations forming the
ash stratum is unknown. No grave goods accompanied any of the three
burials. Details on these burials are as follows:

SURVEY AND EXCAVATIONS ON SAIPAN 55

Burial 1: Shallow, extended burial, lying along the face of the rock shel-
ter, in squares 24CC-24DD. Shoulders of burial were to northeast. No
skull present. All bones in extremely fragmentary condition. No evidence
of cremation. Surface depth to top of burial 1.0 foot.

SECTION XxX!"

= ESP ORAESTH 7
a pe eat
te

Aree eat 7VA x

PLE

Dark gray topsoil

as Ash deposit

[| Limestone outcrop

//) Yellow brown sandy soil
Beach sand

Brown subsoil, sterile of artifacts

Fic. 20. Soil section, Laulau Rock Shelter, Saipan.

Burial 2: Secondary burial lying at the bottom of the ash level in square
24BB. Burial consisted of collection of long bones and phalanges lying over
skull, with one extra mandible among the long bones. Bones in very poor
condition. No evidence of cremation. Surface depth to top of burial 1.2
feet.

Burial 10: Child burial, fully flexed, lying on right side, with skull to-
ward north. Located in square 20AA. Bones extremely fragile, with skull
crushed flat. Surface depth to top of burial 0.7 foot.

56 MARIANAS PREHISTORY

Fic. 21. Burial no. 6, Laulau Rock Shelter, Saipan. Burial lies in pit, the bottom
of which was covered with beach sand.

The remaining twelve burials were all associated with the occupation
level lying under the ash stratum. These burials were all found in pits that
were dug into sterile soil from the light brown stratum above. In no case
were these burial pits intrusive from the ash level, and there was a clear
demarcation between this set of burials and those lying above them. In
some cases, the pits overlapped one another, with one burial superimposed
over part of another.

Of these twelve burials, eight were found lying extended on the back in
pits from 7.0—7.8 feet long and approximately 2.5 feet wide. One burial
had been partially cremated, but the others showed no evidence of fire. All
twelve burials had been placed on a layer of coral beach sand spread in the
bottom of each pit. Bones were generally in fragile condition. No grave

SURVEY AND EXCAVATIONS ON SAIPAN 57

goods were found. The burials were extended parallel to the rock face of
the shelter; otherwise there was no set orientation. Depth from surface to
top of burials ranged from 3.0—3.8 feet, except for one burial] at a depth of
2.4 feet. A typical extended burial is illustrated (fig. 21).

In addition to these eight extended burials, three secondary burials were
found. Each consisted of a miscellaneous collection of bones, in no case rep-
resenting a complete skeleton. Burial 7 was located in a pit sunk directly
over an extended burial (no. 8). It consisted of a skull and fragmentary
long bones, all heavily burned, so the burial must have consisted of bones
remaining from a cremation. Burials 11 and 12 were in pits, with Burial 12
also the remains of a cremation. Of the three, only Burial 12 exhibited a
sand layer that had been spread in the bottom of the pit. No grave goods
were found associated with any of these three burials. Depth of the burials
below present ground surface varied from 2.6 to 2.9 feet.

The remaining burial (no. 5) was probably extended. Only the skull was
uncovered, projecting into the excavated area, and the outline of the burial
pit discerned. Time did not permit excavation of the complete burial.

Stratigraphy

The stratigraphy found at Laulau Rock Shelter may be recapitulated as
follows:

1. An upper ash stratum of old cremations superimposed on an under-
lying occupation level. The lower occupation level exhibited a sandy sur-
face, probably resulting from beach sand spread over the area. Below this
surface were found twelve burials, lying in pits associated with the lower
occupation level.

2. In the upper ash level, the prevailing mode of dispersal of the dead
was obviously cremation, though one secondary burial, one extended
burial, and one flexed burial were found in the cremation ashes.

3. In the lower stratum, the prevailing burial practice was to place the
dead extended on the back in a long pit, the bottom of which was spread
with beach sand.

4. Marianas Plain pottery was found in the ash level. Marianas Red and
lime-filled, impressed sherds, together with some Marianas Plain, were
found in the lower level. No Marianas Red or lime-filled, impressed sherds
were found in the ash level. Further details on the pottery seriation from
this site are given in the chapter on pottery.

5. It should be noted that no burials were found during the excavation
of Laulau House A. It is possible that the occupants of the Laulau latte
houses utilized the rock shelter for disposal of their dead, represented by
the cremations in the upper level of the shelter. If this is so, we do not know

58 MARIANAS PREHISTORY

where the individuals represented by the lower level burials at the rock
shelter lived and built their houses.

EXCAVATIONS AT OLEAI AND CHALAN KIJA

The most favorable geographic location on Saipan for human settle-
ment is along the west coast, fronting the lagoon. In ancient days, here was
fertile land combined with fine fishing, as well as a sheltered lagoon giving
access to the open sea. The earliest voyagers to the island would naturally
have chosen this location.

It has also been the area of the island’s historic settlements: Garapan,
Tanapag, and Chalan Kanoa. Likewise, along the west coast there was a
great concentration of World War II base construction. Unfortunately for
the archaeologist, these last chapters in the human occupation of the island
have destroyed the earlier chapters and most of the archaeological sites
have been obliterated. Careful search did reveal several sites along the west
coast that had not been heavily disturbed, though they were south of what
was once the most fertile agricultural area. Three of these—Oleai, Chalan
Kija, and Chalan Piao—were selected as giving the greatest promise of
finding occupations earlier than those represented by Objan and Laulau.
The first two proved disappointing, but are reported on as a matter of
record.

OLEAI

Description of the Site

Oleai Site is located 300 feet west of the present Oleai village and ap-
proximately 570 feet inland from the lagoon shore (fig. 22). The terrain at
this point is level. Before the war, the land was cultivated and most of it is
still free of bush (fig. 23). The principal surface feature of the site is the
great quantity of shells strewn over an area approximately 500 feet long
and 250 feet wide. Evidently Oleai was at one time a large camp site. There
was no evidence of latte stones.

The name “‘Oleai”’ is derived from a settlement of Caroline Islanders
from Oleai who lived on Saipan for a time during the German regime, after
a typhoon had devastated their home island. Careful inquiry among the
owners of the land made it certain, however, that the archaeological site
was not the result of this late settlement in historic times.

Excavation Procedure and Structure of the Site

Four test trenches, each 5.0 feet square, were excavated along the long
axis of the site. Each trench was taken down in 0.5 foot levels to a depth of
5.0 feet below the present ground surface. At this depth, there was no indi-
cation of cultural material. In none of the trenches were there found any

NG need

wNUOC7, alls,
NUON SG
attlys, Teak c® Shine av\its,
Dic aN00Ss
Test o®
50 ie) 50 100
A oe A ee |
[ === 1
BEES

Fic. 22. Plan of excavations at Oleai Site, Saipan.

60 MARIANAS PREHISTORY

remains of house structures or other architectural features. The site showed
evidence of modern tilling to a depth of 0.5 foot.

The soil structure of all the trenches was identical. The top stratum con-
sisted of a dark gray, humus-filled, sandy soil, varying in thickness from
1.0-1.8 feet. This stratum contained a great many shells, a considerable
number of sherds, and a few other artifacts. A soil profile of Trench D is
shown in figure 23. Below the upper stratum lay light-colored, yellow
beach sand, with a sharp separation in color between the two strata.

All artifacts were recovered from the topsoil stratum. In Trench C a
round, fire-burned area, 1.6 feet in diameter, was found at the bottom
of the upper soil stratum. It extended to a surface depth of 2.0 feet.
Evidently this marked the remains of a cooking fire. It is doubtful that
Oleai Site has any considerable antiquity or that it was long inhabited.

CHALAN KijA
Description of the Site

The Chalan Kija Site is northwest of Lake Susupe, between two marshy
arms of the lake. Here there is a ridge of higher ground about 4.0 feet above
the water table (fig. 24). Surface features of the site consisted of a shell and
sherd area approximately 200 feet in north-south dimension and 75 feet
wide. A few broken /atte stones have been pulled off the site to the edge of
the marsh. In Japanese times, the area was leased to Okinawan farmers
and is today used by Saipan Carolinian families for sweet potato (camote)
fields.

Excavation Procedure and Structure of the Site

Four test trenches were excavated in the most promising sections of the
site and were carried down 3.0 feet into undisturbed, light brown sand.
Trench A was 4.0 X 8.0 feet; the remaining three were 4.0 X 6.0 feet.

The soil structure was the same in all trenches and was similar to that
found at Oleai. The top stratum was a dark gray, sandy soil. Below this was
undisturbed, light brown sand. All shells, sherds, and artifacts were found
in the upper level. No evidence of any feature of house construction was
found in the excavations. Trench C was sunk in a particularly heavy con-
centration of shells.

Chalan Kija Site represented a single and probably relatively brief occu-
pation. The pottery was Marianas Plain, characteristic of latte sites.

EXCAVATIONS AT CHALAN PIAO
DESCRIPTION OF THE SITE

The Chalan Piao Site lies south of Oleai and Chalan Kija, but on the
same level, coastal formation of indurated sand beds. Chalan Piao is lo-

Pe nt

Upper: view of site, looking south; road from beach to

Fic. 23. Oleai Site, Saipan.
village in foreground, sand piles from excavations at Test Trenches C and D in center

background. Lower: soil profile of Test Trench D, showing heavy concentration of

shells and humus in top stratum, underlain with beach sand.

MAG

Pee) )/ 7 ee oa a \Heavy Shell
Concentration «
\ Ae

NUUI/A

Camote Field

|
|
|
|
|
|

¥ Camote

ae
~

-—_— Limits of Sherd Area

&> Breadfruit

K«k Cosuorina

Fic. 24. Plan of excavations at Chalan Kija Site, Saipan.

SURVEY AND EXCAVATIONS ON SAIPAN 63

MAG

RIGHT OF WAY

PLLIVTEEL PLETE EEL ELLE ELLE ETL LEEE EE PPLE EEE REPL ELE PI
es
Gow

RR
/

TITAN Nua da Deda Q ules debe Vola ty de GateUede ToL opoby ult lobelef elute lelole late Calele ede lobe dedsty dele lobobale\ebel sb alelelide

FARMHOUSE

La. RUINS

Try Tpeparyye

TOMaaen

2 40
Mees R

: 40
\@ ce ee L 4
: —— CRATERS ; my,

FEET tae.

Fic. 25. Plan of excavations at Chalan Piao Site, Saipan. Test trenches are num-
bered, according to grid.

We tetede teeter

cated south of Lake Susupe, between a marshy southern extension from the
lake and the lagoon shore (fig. 5). The site is approximately 2,000 feet east
of the lagoon and extends along a low-lying area that slopes eastward to the
marsh (fig. 25). Paralleling the site on the west is the right of way of the
former Japanese narrow-gauge railroad. A few hundred yards northwest of
the site is an abandoned area that contained American petroleum tanks
during World War II. The site itself has been undisturbed, except for shal-

Fic. 26. Excavations at Chalan Piao Site, Saipan. Upper: view of Trenches 35,
50, and 65, looking southwest. Lower: soil profile of Trench 65. Upper stratum,
one foot in depth, consists of dark gray, humus-filled, sandy soil, underlain with light
yellow-brown sand.

SURVEY AND EXCAVATIONS ON SAIPAN 65

Fic. 27. Oyster shell (Ostrea cucullata Lamarck) from Chalan Piao, Saipan. A car-
bon 14 date of 1527 B.C. + 200 was obtained from this shell by W. F. Libby. Shell
exhibits little or no evidence of rolling.

low tillage practiced in Japanese and present times and for bomb craters
that pitted its surface during the World War II invasion of Saipan. The
land is on the farm of Joaquin Cabrera.

The surface of the site was covered with scattered sherds and shells. This
sherd and shell area extended in a northeast-southwest direction for some
350 feet, and was approximately 150 feet wide. West of the railroad right of
way there was also a sherd and shell area. No /atte stones were found at the
site, though the surface sherds were predominantly Marianas Plain, which
is characteristic of /atte sites on the island.

EXCAVATION PROCEDURE AND STRUCTURE OF THE SITE

We excavated eight test trenches at the site, each trench 5.0 feet square
(fig. 25). The soil profiles on these trenches were similar to those at Oleai,
with an upper stratum from 0.08 to 1.0 foot thick, consisting of dark gray,
humus-filled, sandy soil containing sherds and numerous shells. Below this
stratum lay lime-sand, light yellow-brown in color (fig. 26). The first
trenches to be excavated were nos. 35, 50, 65, and 50CC. Unlike Oleai and
Chalan Kija, and to the joy of the digging crew, cultural material in these
trenches was not confined to the upper level; sherds continued to appear as
the excavations were carried down into the lower stratum. At a depth of
2.5—3.0 feet, this sand had become hard-packed, requiring the use of picks.
By a depth of 4.0 feet, the soil was hard-pan, and crowbars had to be added
to the picks. Sherds continued to be found, imbedded in the rock-like
chunks of indurated sand removed from the trenches. The lowest sherd

66 MARIANAS PREHISTORY

(Marianas Red) was found in Trench 125AA at a surface depth of 5.6 feet.
The water table was found at a depth of 5.9 feet, and excavations in
Trenches 35, 50, 65, 50CC, and 125AA were carried to a depth of 6.1 feet
below the surface. Excavations in the other three trenches were carried
down 4.5—5.5 feet below the surface, at least a foot below the lowest sherd.

No evidence of house structures was found in any of the trenches. The
principal interest of the cultural material was the pottery seriation deter-
mined, with Marianas Plain ware predominating in the upper levels of the
site but giving way to Marianas Red sherds in the lower levels. This seria-
tion is discussed at greater length in the chapter on pottery. Another point
of great interest was the finding of a large oyster shell (Ostrea cucullata
Lamarck) at a depth of 1.5 feet in Trench 125AA (fig. 27). The shell was
subsequently determined by W. F. Libby to have a radiocarbon date of
1527 w.c. + 200 years (Libby, 1952).

There is little doubt that the Chalan Piao Site has a respectable antiq-
uity. From the point of view of age, it proved to be the most important site
excavated during the season.

ANTIQUITY OF THE SITE

The principal problem concerning the antiquity of the Chalan Piao Site
revolves around the question: How did the sherds, found to a depth of 5.6
feet, get there? This in turn raises the question—a strictly geological one—
as to the manner of deposition of the limestone sands in which the sherds
were imbedded.

Two possible methods of deposition have been suggested by Dr. Preston
E. Cloud, of the United States Geological Survey, who is most familiar
with the geology of Saipan. One is that the 2,000 feet of limestone sands
that stretch between the site and the lagoon were built up and outward by
storm action, the sherds having been deposited from camp sites along the
shore. However, in this case one would expect that the sherds would ex-
hibit pronounced evidence of rolling, and most of them exhibit very little.
The oyster shell in particular exhibited none. There is also the matter of
adjusting this explanation to the 6-foot fall in sea level related to the post-
glacial optimum.

The second possibility is that the limestone sands represent a lagunal
deposit, laid down during the post-glacial optimum, and that the sherds
were deposited from a camp site off the toe of a sand spit, formed to the
west of the site proper. The only difficulty with this hypothesis is that the
sherds are not covered with encrusting marine organisms, a point raised by
Dr. Cloud. However, this hypothesis seems more in accord with the evi-
dence.

SURVEY AND EXCAVATIONS ON SAIPAN 67

In any case, the geological evidence is not in opposition to the radio-
carbon date determined from the oyster shell. This oyster was probably
consumed by man, and the shell discarded, along with broken pottery.
The geological evidence tends to support the antiquity of the site as given
by the radiocarbon date.

TT: Survey and Excavations on Tinian

PHYSICAL CHARACTERISTICS OF TINIAN

Tinian is approximately 12 miles long and 6.5 miles across at its widest
point, and has an area of a little over 39 square miles. As in the case of
Saipan, Tinian consists of a series of raised limestone terraces. Bowers
(1954) notes that the island is composed of two major plateaus, separated
by the Marpo Valley. Each plateau is in turn topographically composed of
a set of terrace levels. Elevations on Tinian are lower than on Saipan and
there is relatively more level land. Mount Lasso, the highest elevation on
the island, is only 564 feet.

The coast of Tinian consists largely of limestone cliffs dropping ab-
ruptly into the sea, but this rocky shore line is broken by several beach
areas and small coves. Particularly important is the Tinian harbor area on
the west side of the island. This has always been the principal landing point
on the island and has been a local center of settlement since prehistoric
times.

DESTRUCTION OF ARCHAEOLOGICAL SITES

Tinian was a major American military base during World War II, and
as in the case of Saipan there has been much disturbance of sites, particu-
larly around the harbor area. However, relatively more /atte sites remain
intact than on Saipan. A number of caves were investigated on Tinian, but
all had been converted into military strong points by the Japanese and
were so disturbed that whatever archaeological evidence they may once
have contained has been destroyed.

SURVEY

As in the case of Saipan, the archaeological survey of Tinian was first
concentrated along beaches and coastal terraces. Inland areas were then
explored, and finally attention was paid to caves. In conducting the ar-
chaeological survey, I received invaluable aid from the members of the
United States Geological Survey field team then working on Tinian. By the
time of my arrival, Messrs. Doan, May, and Burke, who composed the
team, had covered almost every square foot of Tinian and had already

|
145° 40'

A USHI PT.
4
4
elk = 4

ASIGA PT.
HILO PT.

EARLE

MASALOG
_ PT.
a Sie 1D
GURGA
PT. :
TINIAN “EE
HARBOR gq} ,
ye : MARPO PT.
and
f
LALO PT.
r— 55
= 30"
L534"
145° 40

Fic. 28. Principal archaeological areas and sites of Tinian. 1, Nuclear Tinian:
(a) Leprosarium I, (b) Leprosarium II, (c) House of Taga, (d) Blue Site. 2, Marpo
Valley: (a) Marpo I, (b) Marpo II, (c) Marpo III. 3, Unai Dangkulo: (a) Unai Dang-
kulo I, (b) Unai Dangkulo IT, (c) Unai Dangkulo III, (d) Unai Dangkulo IV. 4, Sec-
ondary areas: (a) Latte site in Gurgan-Hilo area, (b) Central bomb dump.

70 MARIANAS PREHISTORY

found all the major archaeological sites extant. The information that fol-
lows on the location of Tinian sites is almost entirely due to the co-opera-
tion of these geologists.

Since shortly after the close of World War II, Tinian has been inhab-
ited by a small community of Chamorros, by the Trust Territory Lepro-
sarium, and by a few American administrative personnel. These groups
occupy only a small part of the island. The remainder has been overgrown
with a dense cover of bush, which made survey work arduous and difficult.
However, I feel reasonably certain that most of the sites still existing were
discovered and located.

The archaeological sites discovered on Tinian can be grouped into three
major areas, together with minor, less important sites (fig. 28). These major
areas are:

1. NUCLEAR TINIAN

This area extends from a point slightly north of the present Tinian
leprosarium to the Blue Site on the south. The area includes the harbor
shore, the natural geographic location for settlement. Major sites discov-
ered in this nuclear area are:

(a) Leprosarium I: The present (1950) leprosarium occupies an old
archaeological site which extends along a terrace to a small point jutting
into the sea northwest of the present leprosarium buildings. The area is at
least 400 yards in length. The northwest section of the area, about five
acres in extent, has been used for small farms by the leprosarium patients.
Everywhere in this area there is a heavy concentration of surface sherds.
One badly disintegrated 8-shaft latte house still survives and numerous
shaft and capstone fragments are scattered over the area. There is one well-
defined low mound, either a house site or a shallow refuse deposit. There is
no doubt that this area was once a major site, but it has been much dis-
turbed.

(b) Leprosarium II: North of Leprosarium I is a sherd area approxi-
mately 50 yards square and one 8-shaft latte house. The shafts and capstone
fragments were greatly weathered. The long dimension of the house paral-
lels the coast line.

(c) House of Taga: This is the center of nuclear Tinian. The House of
Taga is an immense 12-shaft latte structure, preserved in a small park. It is
all that remains of a prehistoric village, found by Hornbostel in 1924 to
consist of 18 separate /atte structures. The House of Taga is described in
detail in a following section.

(d) Taga Quarry: Located on the shore approximately 4,000 feet south
of the House of Taga. Details are given in a following section.

SURVEY AND EXCAVATIONS ON TINIAN va

(e) Blue Site: An extensive series of latte houses at the south margin of
the harbor area. Details are given in a following section.

2. MARPO VALLEY

The interior of Tinian is occupied by the Marpo Valley. This valley has
two arms, one extending north-south and the other southwest-northeast.
The two arms meet in a depressed area, the lowest part of which is occupied
by a small swamp. The soil of the valley is fertile, and the small swamp
supplies fresh surface water. Both the Japanese and the American military
forces installed pumping stations at this site to provide other parts of the
island with fresh water.

The Marpo Valley is a major archaeological area. In both arms of the
valley surface sherd areas are common, though deposits of cultural ma-
terial everywhere appear to be shallow. Marpo is known to have been
occupied by Chamorros at the time of Spanish discovery. One of the early
Spanish accounts mentions that Sanvitores, who established the Jesuit mis-
sion in the Marianas, in 1699 settled a local civil war on Tinian between
the residents of Marpo and those of Sunharon, the latter name applying to
the harbor area.

During the survey the following Jatte sites were located in the Valley:

(a) Marpo I: In the southeast corner of the valley, just east of the
swampy area, a single 8-shaft latte was found. House dimensions were
12 X 38.5 feet. The site was located on a former Japanese farm and was
considerably disturbed. Shafts were much weathered and tops were bro-
ken, except for one which had fallen and remained relatively intact. This
shaft was 3.5 feet long, 1.4 X 2.6 feet at the base, and 0.9 X 1.3 feet at the
top. No capstones were found intact. One well-preserved mortar was found
associated with the house. The long dimension of the house was oriented
north-south, at right angles to the slope of the ground. Approximately 50
feet north of the house several fragments of additional shafts were found,
indicating the probable former presence of another /atte house.

(b) Marpo II: Southwest of the American-built pumping station and
close by the road is a single, large latte house. Only five shafts project above
the surface, though the structure probably contained eight shafts. Existing
shafts show considerable weathering. No capstones were found intact. Di-
mensions of the house were 12 X 44 feet. Orientation was north-south, at
right angles to the slope of the ground. The best-preserved shaft projects
5.1 feet above the present ground surface. Dimensions of this shaft at
ground surface were 1.9 X 3.0 feet; at top, 1.4 X 1.5 feet. The shafts at
Marpo II, as at most other Tinian sites, are more trapezoidal in form than
those of Saipan.

We. MARIANAS PREHISTORY

(c) Marpo III: Consists of two latte houses, located in the north arm of
the valley on its east slope, and several hundred yards east of the road.
The two structures were built end to end, separated by a distance of 40 feet.
The northernmost house was a 10-shaft /atte, 13.5 & 45 feet. Six shafts are
erect and in place, but they are all much weathered. The best-preserved
shaft projected 3.0 feet above the ground. Dimensions of this shaft at
ground surface were 2.3 X 2.7 feet; at top, 1.1 X 1.2 feet. Stone mortar
was found at the southwest corner of the house. The other house had only
three broken shafts in place and its original dimensions could not be deter-
mined. Fragments of other shafts and of capstones are in the immediate
vicinity. The two houses were oriented north-south, with long dimension at
right angles to the slope of the ground.

Below Marpo III and somewhat to the south in a cultivated field, six
circular dark-colored areas, approximately 30 feet in diameter, were also
found, together with surface sherds. These may have been either house or
burial sites, though the cultural deposit appeared very shallow.

3. UNAI DANGKULO

In the bay formed between Asiga and Masalog Points on the east coast
of Tinian there extends a stretch of beach and fringing reef that is similar
to Laulau beach on Saipan’s Magicienne Bay. This beach area on Tinian
is known as Unai Dangkulo (‘“‘big beach’’). Around this beach is a concen-
tration of latte sites. The location combines good reef fishing with adequate
farmland, a combination that formed the basis for the location of almost
all coastal /atte sites in the Marianas.

At the time of the Tinian survey, the Unai Dangkulo area was so heavily
overgrown with bush that I am not at all certain that all sites were dis-
covered.

(a) Unai Dangkulo I: A small, very badly disturbed, 8-shaft latte. Two of
the interior shafts were missing and the others were so badly weathered
that no accurate dimensions were possible. No capstones were found intact.
House dimensions were 12 X 31 feet. The structure was oriented parallel
to the coast line. For approximately 100 feet on either side of the house was
a moderate concentration of surface sherds.

(b) Unai Dangkulo II: A large 12-shaft latte, 15 X 60 feet. All shafts
were standing, with fallen capstones. A few feet to the south of this atte was
a badly disintegrated 8-shaft latte. This site was literally buried in bush
containing dozens of wasps’ nests. The occupants of the latter found the
intruding archaeologist an easy target and survey operations at this site
were brought to an untimely close. It is probable that additional /atte exist
at this site. The two that were discovered were oriented with the long
dimension parallel to the coast line.

SURVEY AND EXCAVATIONS ON TINIAN 73

(c) Unai Dangkulo III: Consists of three latte structures, with possibly
others whose shafts have disintegrated, found in an abandoned sugar cane
field. The largest of the three was a 12-shaft /atte, with dimensions approxi-
mately 14 X 58 feet. Only six badly weathered and broken shafts remained
in place, but fragments of other shafts and of capstones were scattered
about. Fifty feet north was found a 10-shaft latte. The shafts of this house
were in place but also very much weathered and broken. Dimensions of
this house were 14 X 44 feet. Approximately 50 feet north of this second
latte was a third. The third was probably an 8-shaft structure, but only a
few badly weathered and broken shafts remained in place. It is probable
that other /atte structures once existed south of the large 12-shaft atte. The
entire area of this site was covered with sherds. The three Jatte were ar-
ranged end to end, roughly parallel to the coast line.

(d) Unai Dangkulo IV: North of Unai Dangkulo III the remains of a
single /atte house were found. Only two broken shafts were found in place,
with the disintegrated fragments of others at the site. One stone mortar was
found associated with the latte.

4. SITES OF SECONDARY IMPORTANCE.

(a) Along the west coast of Tinian, particularly between Gurgan Point
and Hilo Point, is a coastal terrace of relatively fertile though shallow soil.
Along this terrace extensive, but intermittent, sherd areas are to be found.
In his field notes, Hornbostel grouped this area with what I have called
“nuclear Tinian’ into a single major archaeological area. Primarily the
Gurgan-Hilo region consists of surface sherd areas, but one Jatte site was
found. The location of this site is shown in figure 28. It consists of two
probable 10-shaft and two probable 8-shaft /atte, but the shafts are so
broken and weathered that accurate dimensions could not be obtained.
These /atte are aligned end to end from 35 to 100 feet apart and parallel to
the coast. Two additional latte probably existed to the south, judging from
the number of limestone fragments. Two stone mortars were found.

(b) In the north central part of the island, an extensive sherd area, to-
gether with fragments of /atte shafts and capstones, was found. This area
was the site of a large World War II bomb depot. At the end of the war,
the bombs in the revetments were blown in place, leaving tremendous
craters. Consequently only bits and patches of the archaeological site re-
main, where the original ground surface still exists.

EXCAVATIONS AT THE BLUE SITE

One of the objectives of the archaeological survey on Saipan and Tinian
was to find an entire /atte village relatively intact. Although it is very prob-

74 MARIANAS PREHISTORY

able that many prehistoric Chamorro villages consisted largely of wood
structures, previous survey work by Hornbostel had revealed clusters of
stone-pillared house sites at Agingan on Saipan and at the Taga Site on
Tinian (Thompson, 1932, pp. 17-18). At the time of my own work, the
Agingan Site and all except the central /atte structure at the Taga Site had
been destroyed. The Blue Site was important in that it was the only major
site discovered where an aggregation of stone /atte houses was found intact.

The Blue Site derived its name from the fact that it lies just back of what
American invasion forces in 1944 designated as Blue Beach. There is no
surviving Chamorro place name for this location, so the wartime designa-
tion has been used.

LOCATION AND GENERAL DESCRIPTION

The Blue Site (see fig. 29) lies along the western edge of an abandoned
sugar cane field. The area between the site and the beach and over half of
the site itself were heavily overgrown with a dense cover of trees and bush.
During the 1944 battle for Tinian the area was bombarded, but there was
little wartime destruction of the site. There was no landing by American
forces at this point.

The Blue Site consists of 10 /atte structures, strung roughly end to end in
an uneven line that parallels the beach (see fig. 30). The site is well located,
in that there is good inshore fishing along the beach and shallow but good
soil for farming inland from the site. Around the lattes themselves there was
a heavy surface concentration of shells and sherds. The sherd area extended
west of the Jatte about 50 feet and inland across the road (fig. 30).

A number of small, low mounds were found along the seaward side of
the Jatte structures. Here the ground commences to slope toward the beach,
though the latte area itself is level. Test trenches were sunk into two of these
mounds and both proved to be refuse dumps. It is possible that the other
low mounds are also the result of refuse accumulations.

Of the 10 Jatte structures, the largest and most imposing was at the cen-
ter. This arrangement also held true at the Taga Site (Thompson, 1932,
pp. 17-18). Whether the large central latte served some special function,
either as a chief’s house or as a ceremonial structure of some sort, could not
be determined. The central atte had all the remains of daily life—sherds,
shells of edible mollusks, adze blades, etc.—on and in the ground about it,
so that it must have been used as a living house and not reserved for special
occasions. However, the village plan, with the largest house at the center,
may well be a reflection of the social organization of the community.

In figure 30 the latte structures are numbered from I to X. They were all
badly weathered, and in no case were the stone capstones found in place
on the shafts. It is quite possible that there were wood houses also at this

SITE OF TINIAN TOWN A
N

— (DESTROYED)
TAIN ele

HOUSE 7.
OF TAGA .

1000 QO 500 1000
——

FEES }.
BLUE SITE LOCATED 1

RESPECT TO THIS CORNER
(SEE MAP OF BLUE SITE)

Fic. 29. Location of Blue Site, House of Taga, and Taga Quarry, Tinian.

76 MARIANAS PREHISTORY

site and that these have long ago disintegrated, and also there may have
been a few additional stone Jatte houses that have disappeared, leaving no
surface indication.

Blue I was the only house site excavated. Brief descriptions of the remain-
ing latte follow. All shafts and capstones were cut from coral limestone.

Blue II, an 8-shaft latte. Only one shaft had not broken off, though all
bases of other shafts were in place. The intact shaft projected 3.0 feet above
the present ground surface and was trapezoidal in form. One intact cap-
stone and two fragmentary capstones lay near the shaft bases, as well as one
large stone mortar. Dimensions of structure: 36 X 12 feet.

Blue III. Probably an 8-shaft latte, though only the bases of five shafts
projected above ground. All shafts were very much eroded. No capstones
were found. Two stone mortars were found 25 feet west of the house. Prob-
able dimensions of structure: 30 X 12 feet.

Blue IV, a 10-shaft latte, in fairly good condition. Five shafts were intact,
all of trapezoidal form. These shafts projected above the present ground
surface from 3.4 to 3.9 feet. Eight capstones were found, all of typical cup-
shaped form. Dimensions of the best-preserved capstone were 3.5 feet
(upper surface diameter) X 1.8 feet (lower surface diameter) X 2.0 feet
(thickness). A fragment of stone mortar was found at the east side of the
house. Dimensions of structure: 41 X 13 feet.

Blue V. Probably an 8-shaft Jatte, but two northernmost shafts had dis-
integrated. The remaining shafts were broken and greatly weathered. One
shaft base projected 2.8 feet above the present ground surface, but origi-
nally it undoubtedly had a greater height. Only one weathered capstone
fragment was found. Probable dimensions of structure: 36 X 12 feet.

Blue VI. Probably an 8-shaft latte. Though only three shafts were evident,
two of these were at one end of the house and one at the other, with two
fallen capstones regularly spaced between the ends. The two shafts at one
end had fallen outward, but they were in fairly good condition. Both were
trapezoidal in form, and from 3.6 to 3.7 feet in length. Of the third shaft
only the base remained. Six capstones were found. Dimensions of the best-
preserved one were 3.2 feet (upper surface diameter) X 1.7 feet (lower
surface diameter) X 2.1 feet (thickness). Probable dimensions of structure:
35 X 14 feet.

Blue VII, Probably an 8-shaft latte, but only bases of three shafts were
found in place, together with fragments of two capstones. The structure
was located in a sugar cane field, and most of the capstone and shaft frag-
ments apparently had been removed. Probable dimensions of structure:
34: XX 11 feet:

Blue VII, an 8-shaft latte. One interior shaft was missing; other shafts
were broken or eroded, but the bases were found in place. The best-pre-

‘uelury, “ag ong Jo dep “o¢ “Oly

4334
ES REE 5] OTSA FIN OOO EUR OSINS. CO. FEN tOVO NEY-E2
00! os te) os
| Y3NYOD O1 1334 ved Stent
“+
O7F/4 Lone Q737!1J INVID HVINS GAINOUGNVEGY
oo
INVI wvgans eee” mx
GINOONVEY er Nes Ponies ee I ie eee oe
Zw === Se A Sore aN
xr Cs a ak af wiahs Sr Clare Eee Sa 5 tae ee eae ie
De Pane ae ai: DT ree
se: & oe ti V 1831) Bae) oie
ice ele sees YBLVYO TIBHSO, © g sai?” = oeoe Ix
eee QNNOW Catia ate we
oe ee eae cet 3SN33y GNNOW miata
asni3uy
HSNG@ IJSNIO -SIIHL TIVWS HSNG@ FISNIO - SFFVL TIVWS
LOSES ane opecpuns AOVLE .
vais FN edo AD Tahoe

Td.

78 MARIANAS PREHISTORY

served shaft, though eroded, was trapezoidal in form and projected 3.1 feet
above the present ground surface. One capstone fragment was found.
Four feet from the southwest corner of the house was a well-made stone
mortar. Dimensions of structure: 32 X 13 feet.

Blue IX, an 8-shaft latte. One interior and one corner shaft were missing,
but fallen capstones marked their former location. Of the six remaining
shafts, four were weathered but nearly intact; of the other two only the
bases remained in place. The four nearly intact shafts were trapezoidal and
projected from 2.6 to 3.0 feet above the present ground level. Seven fallen
capstones were found. Their dimensions varied from 2.9 to 3.2 feet (upper
surface diameter) X 1.4 to 1.6 feet (lower surface diameter) X 1.5 to 2.0
feet (thickness). Dimensions of structure: 33 X 12 feet.

Blue X, an 8-shaft latte. One interior shaft was missing; the other shafts
were broken and weathered, but the shaft bases were found in place.
Three capstones were found, only one of which was intact. Its dimensions
were 3.0 feet (upper surface diameter) X 1.5 feet (lower surface diame-
ter) X 1.8 feet (thickness). Dimensions of structure: 31 X 12 feet.

Excavations AT BLuE I
Excavation Procedure and Structure of the Site

Blue I, the largest of all Blue Site atte, was a 12-shaft latte structure, 58
feet long and 14 feet wide, located at the center of the line of /atte that
formed the site. A ground plan is shown in figure 31.

The surface of Blue I was heavily covered with potsherds, with both
whole shells and fragments of Tridacna, as well as of other shells, and with
some stone tools and flakes of volcanic rock. The latte was heavily over-
grown with bush, though at one time the house area had been shallowly
plowed. A stone mortar was found approximately 55 feet west of the house.

The excavations at Blue I consisted of a trench, 5.0 feet wide and 45.0
feet long, that was cut across the center of the latte at right angles to its
long dimension. The soil proved to be shallow, and the trench was taken
down to the underlying limestone bedrock. The surface of this bedrock was
very uneven. The two deepest depressions were only 2.1 and 2.2 feet below
the present ground surface. In some spots the limestone was found within
0.4 foot below the surface; its average depth was 1.2 feet (see figs. 32, 33).

Lying in the bottom of depressions and pits in the bedrock were found a
few areas of undisturbed, red-brown soil, never more than 0.1 foot thick.
For the rest, the entire soil stratum was loose and friable, dark gray in
color, and filled with sherds, shells, stones, remains of burials, and charcoal.

No evidence was found of a coral pebble stratum, as was discovered at
Objan and Laulau sites on Saipan. Nor were there found any remains of

Q) MAG

32 FEET E ’
To TEST A pl
6 Ee
VE)
5 8

x 10
N a
2 ead ||

fe) bo ie) 5 fe) 20
_——— 252-2 —SSS=S=
| 2

Fic. 31. Plan of Blue I, Blue Site, Tinian. Latte shafts numbered; capstones let-

tered; limits of excavation indicated by broken lines.

80 MARIANAS PREHISTORY

an earth oven or a cook house in the inland extension of the trench, though
the soil was filled with minute charcoal fragments and considerable ash.

In addition to the lateral trench excavated across the house, test trenches
were excavated in two small refuse dumps located to the seaward side of
Blue I (see fig. 30).

SECTION THROUGH SHAFTS 4 AND 9

10 feet
a

9h Coral limestone

Dark gray topsoil

Fic. 32. Section through shafts 4 and 9, Blue I, Blue Site, Tinian.

Test Trench A~—Excavated in refuse mound to the northwest of Blue I.
This mound was approximately 34 feet across in east-west dimension and
25 feet in north-south dimension. A World War II shell crater, approxi-
mately 8 feet in diameter, was located on the northern margin of the
mound. The test trench, 5.0 feet wide and 10.0 feet long, was sunk into the
center of the mound and excavated down to the limestone bedrock, which
lay from 3.1 to 4.0 feet below the present ground surface. Except for a
lower thin stratum of red-brown, sterile soil, from 0.1 to 0.4 foot thick, the
entire deposit consisted of an accumulation of ashes, broken shells, sherds,
and miscellaneous debris, capped with a thin recent deposit of topsoil.
Lenses of lighter- and darker-colored ash appeared on the soil profiles
along the sides of the trench, indicating successive dumpings of ashes prob-
ably removed from cooking hearths. At the bottom of the trench was found
a large accumulation of fish scales.

Test Trench B.—Excavated in refuse mound to the southwest of Blue I.
The trench was 5 feet square and was carried down to bedrock, which was
found to lie from 3.5 to 3.7 feet below the present surface. The soil structure

Fic. 33. Blue I, Blue Site, Tinian. Upper: site cleared of bush and staked, show-
ing /atte shafts; looking north, with shafts 4 and 9 in foreground. Lower: trench, 5 feet
wide, excavated across site, showing shallowness of soil overlying rough-surfaced, coral
limestone bedrock; looking west.

82 MARIANAS PREHISTORY

Fic. 34.. Shaft 7, Blue I, Blue Site, Tinian. Note markedly trapezoidal form of
shaft, which projects 5.7 feet above present ground surface.

was the same as that found in Test Trench A—a thin lower stratum of red-
brown soil, above which lay the refuse deposit, capped with a thin layer of
topsoil. The refuse deposit consisted largely of ashes, with numerous sherds.

Latte

SHAFTS

Weathering.—Latte shafts exhibited great variation in amount of weather-
ing and degree of disintegration. Shafts 3, 4, 6, 7, 9, and 12 were largely
intact; other shafts had broken off, though bases remained in place (fig.
33)s

Material.—All shafts were coral limestone.

SURVEY AND EXCAVATIONS ON TINIAN 83

Form and Dimensions.—Shaft 7 was the best preserved and is illustrated
(fig. 34). Blue I shafts are more trapezoidal, with broader bases and nar-
rower tops, than those found on Saipan. Eliminating shaft 3, whose top
was badly weathered, the remaining standing shafts projected from 5.15 to
5.75 feet above the present ground surface. Shaft 7 measured 1.6 X 1.5
feet at the top and 4.1 X 2.5 feet at the point where the base entered the
ground. The tops of the shafts were approximately level with one another.

Constructional Features—Exploratory trenches were excavated around
shafts 4 and 7. Both these shafts were found to rest on the limestone bed-
rock, with rocks packed around the base of each shaft to keep it upright.

CAPSTONES

Weathering.—No whole capstones were found, though numerous frag-
ments of disintegrated capstones lay near the shafts.

Material.—All capstone fragments were coral limestone.

Form and Dimensions.—All capstone fragments displayed the usual cup-
shaped form. From the fragments, it was possible to estimate that the cap-
stones were from 1.8 to 2.0 feet in thickness and that their Breeton diameter
was approximately 4.0 feet.

Burials

In the trench excavated across the /atte structure, three burials were
found, and in addition a miscellaneous assortment of human bones not
associated with well-defined burials. These were all located west (seaward)
of the center of the house. Probably the entire western half of the unexca-
vated house area was similarly used for burials.

Burial 1: Secondary adult burial, consisting primarily of skull and long
bones. Bones present in good condition, but many bones of the skeleton
missing. No grave goods. Depth to top of burial 1.1 feet (see fig. 35).

Burial 2: Multiple secondary bundle burial, consisting of two skulls and
a miscellaneous assortment of long bones and a few phalanges. Bones in
fair condition. No grave goods. Depth to top of burial 1.05 feet (see fig. 35).

Burial 3: The upper half of what was probably an extended burial.
Bones present, all in poor condition, were those of skull, thorax, and arms.
Burial was oriented on back in east-west direction, with skull at east. A
few long bones were found where the lower half of the burial would nor-
mally have been, but these long bones probably represent a later secondary
burial of a few human bones. No grave goods found. Depth to top of burial
1.8 fect:

All these burials were lying on the surface of the limestone bedrock at
the bottom of the trench.

i : i : Burial 1.
Fic. 35. Burials, Blue I, Blue Site, Tinian. Upper: Burial 2. Lower: Bu

SURVEY AND EXCAVATIONS ON TINIAN 85

Pathology Indicative of Yaws

One skull and certain of the long bones of Burial 2 are of particular in-
terest because they exhibit pathology characteristic of yaws and at the
present time are the earliest evidence of yaws that exists in the records of
paleopathology. These pathological bones have been examined by T. Dale
Stewart and his observations and conclusions are recorded in a separate
paper (Stewart and Spoehr, 1952).

Radiocarbon Dating of Blue I

Both test trenches at Blue I yielded liberal numbers of Tridacna shells
among the refuse of which the mounds were composed. A Tridacna shell
found at a depth of 1.9 feet in Trench A was dated by the radiocarbon
method at A.D. 845 + 145. Inasmuch as the refuse mound from which this
Tridacna shell came was clearly associated with the Blue I /atte, and as the
deposit of cultural material at this site was so shallow and hence indicated
only a relatively brief period of occupation, the radiocarbon date provides
a carbon 14 date for the Blue I Site. The 7ridacna is an edible mollusk and
the shell supplying the radiocarbon date was undoubtedly deposited on the
refuse dump shortly after the animal was consumed by man.

EXCAVATIONS AT: THE TAGA SITE
LocATION AND GENERAL DESCRIPTION

The House of Taga, consisting of twelve immense shafts and capstones,
is located at the present harbor area (fig. 29). Originally, the site was ap-
proximately 360 feet east of the shore, but with the construction of World
War II harbor improvements the beach area was filled in and the shore
line moved farther west.

When Hornbostel visited the Taga Site in 1924, he found that it con-
sisted of eighteen separate latte structures, extending roughly end to end
and parallel to the beach (Thompson, 1932, p. 18). The House of Taga
was the largest of these structures and was located near the center of the
line. The site was composed, therefore, of an entire assemblage of latte
houses, very similar to the present plan of the Blue Site, but more exten-
sive. Also, as in the case of the Blue Site, the most impressive house was
built near the center of the line.

After the time of Hornbostel’s visit, the Japanese developed Tinian for
the production of sugar cane. The town of Tinian grew up around the
Taga Site, as the location at the harbor was the natural spot for a modern
population to settle. All the lattes except the House of Taga were removed
to make way for streets and buildings. The House of Taga and the ground
immediately surrounding it, however, were made into a small park. Trees

86 MARIANAS PREHISTORY

were planted around the margins of the park and the interior area was
planted to turf.

In 1944, Tinian was invaded by American forces during the Marianas
campaign of World War II. Tinian town was completely destroyed, and
the harbor area was bulldozed and rebuilt as a military area. The House of
Taga was fortunately set aside and surrounded by barbed wire, and hence
escaped the attention of American bulldozers. By 1950, a certain amount
of vegetation had grown on and around the shafts and capstones, but they
remained in good condition.

THE TAGA SITE IN HIsTORY

I have found no references to the Taga Site in the early Spanish ac-
counts but with the publication of Lord Anson’s story of his voyage around
the world in the years 1740-44, Tinian became known in detail to the Eu-
ropean world. Anson arrived at Tinian in August, 1742, with his crew in
deplorable condition from the ravages of scurvy. Except for a small forag-
ing party of Chamorros and one Spaniard, the island was uninhabited, but
it was well stocked with provisions in their natural state, and Anson and
his men quickly recovered their health. Anson’s account is one of the best
concerning the island. He notes the large number of archaeological sites:
‘‘For there are, in all parts of the island, a great number of ruins of a very
particular kind; they usually consist of two rows of square, pyramidal
pillars, each pillar being about six feet from the next, and the distance be-
tween the rows being about twelve feet; the pillars themselves are about
five feet square at the base, and about thirteen feet high; and on top of each
of them there is a semi-globe, with the flat part upwards... .”

Anson does not specifically state that he is here describing the House of
Taga, but the large dimensions of the latte given in this quotation could
refer only to the Taga Jatte. He also includes a view of his landing and
watering place at Tinian harbor (fig. 2). In this view the Taga Jattes are
prominently featured and he refers the reader of his archaeological de-
scription to this illustration. It is interesting to note that all twelve shafts
of the House of Taga are pictured as still standing, with their capstones in
place.

Anson also notes that the small party of Chamorros that he captured on
Tinian assured him that the Jatte found in various parts of the island were
the foundations of buildings. Anson’s visit was less than fifty years after the
inhabitants of Tinian had been conquered by the Spanish and removed
to Guam.

After Anson’s day, the House of Taga appears intermittently in the liter-
ature. Mortimer, in 1791, recorded the presence of the Taga Site, but gave
no extended description. Arago, who was with Freycinet in the Marianas in

400 ft.

VILLAGE

Fic. 36. Map of Taga Site as it was in 1924 (after sketch by Hornbostel). Site
consisted of 18 /atte structures, indicated by numbers on map.

88 MARIANAS PREHISTORY

1818, reports that seven shafts were standing erect. Sanchez y Zayas ex-
amined the site in 1865 and likewise reported seven standing shafts.
Marche (1891) described the House of ‘Taga briefly and included a photo-
graph. Fritz (1904) stated that by 1900 all but five shafts had fallen. When
Hornbostel visited the site in 1924, only two shafts remained erect with
their capstones in place.

Figure 36 reproduces Hornbostel’s sketch map of the Taga Site. On the
map are plotted the locations of the eighteen Jatte. Each latte was examined
by Hornbostel, and the following condensed description of the seventeen
latte that have been destroyed is taken from his notes.

Latte 7: 12-shaft latte in good condition. Shafts in place, projecting ap-
proximately 5.0 feet above ground surface. Dimensions of latte 10.9 & 61.6
feet.

Latte 2: 12-shaft latte, badly weathered. Shafts out of place, so no meas-
urements were taken.

Latte 3: 10-shaft latte, with four shafts in good condition; all but two
capstones broken. Shafts projecting approximately 3.6 feet above ground
surface. Base of capstones flat, with a slot cut into it to receive the top of the
shaft. Dimensions of lJatte 11 X 40 feet.

Latte 4: 10-shaft latte, with five shafts intact; others broken but bases in
place. Intact shafts projecting approximately 4.6 feet above ground surface.
All capstones broken. Dimensions of /atte 11 X 42.6 feet.

Latte 5: 12-shaft latte, with all shafts in place, but some broken. Shafts
projecting approximately 3.8 feet above ground surface. Eight capstones
present. Dimensions of latte 12.6 X 54.4 feet.

Latte 6: 8-shaft latte, with shafts and capstones badly broken. No meas-
urements taken.

Latte 7: Probably 12-shaft latte. Seven shafts present, projecting approxi-
mately 2.0 feet above ground. All capstones missing. Probable dimensions
of latte 9 X 35 feet.

Latte 8: Only one shaft and two capstones intact. Shaft projecting 4.0
feet above ground. Other shafts and capstones broken and some removed.
No measurements of house taken.

Latte 9: Only shattered fragments of shafts and capstones found. Prob-
ably a 10-shaft Jatte.

Latte 10: Only one shaft found, with fragments of other shafts and cap-
stones about. Shaft projecting 3.5 feet.

Latte 11: Only one shaft and capstone intact. Shaft projecting 5.3 feet
above ground. Capstone 5.2 feet in diameter.

Latte 12: Only one capstone intact, with other capstones and shafts re-
duced to fragments.

SURVEY AND EXCAVATIONS ON TINIAN 89

Latte 13: One complete and two broken shafts found in place, with one
intact capstone. Other shafts and capstones missing. Intact shaft projecting
3.6 feet above ground.

Latte 14, 15, 16: Hornbostel notes these ‘. . . are all in a ruined condi-
tion and measurements would be of no value. They appear to have been of
8 stones (shafts).”’ No. 14 had three shafts and nos. 15 and 16 each two
shafts intact. Shafts projecting from 2 to 3 feet above ground.

Latte 17: 12-shaft latte, with six shafts standing, six fallen. Standing
shafts projecting approximately 5.4 feet above ground. All capstones in-
tact, but displaced from shafts. Diameter of capstones 5.3 feet. Dimensions
of latte 12.7 X 54.5 feet.

THe TAGA SITE IN CHAMORRO MYTHOLOGY

According to Chamorro legend, Taga was an ancient Chamorro giant,
a man of great exploits, among them being the construction of the struc-
ture on Tinian that now bears his name. The Chamorros of today have
been subject to influences from the West and from Japan for many years,
so that Chamorro folklore only survives in a fragmentary state. The follow-
ing account was obtained from Alberto Tobis, an elderly Chamorro who
was born on Saipan. The account contains incidents also found in the texts
of other stories, as recorded by Gertrude Hornbostel on Guam and pub-
lished in Thompson (1932). The Taga account also ends on an obviously
modern note.

Taga was originally a chief on Guam. He was a very big man, at least ten feet tall.
He came from Guam to Rota, fought the Rota chief, and defeated him. Then Taga
became the chief of Rota. He married a Rota woman and had a girl child by her.
Taga commenced to build himself a house on Rota, and started to quarry the shafts and
capstones at As Nieves. But he never finished the quarrying, as he decided to go to
Tinian instead. So he left Rota and went to Tinian.

On Tinian, Taga met the Tinian chief for a contest of strength. First, the two chiefs
both got in one canoe, sitting back-to-back. Taga faced Rota and the Tinian chief,
Tinian. At a given word they both paddled as hard as they could. The canoe cracked
apart between them. The two chiefs were even on this contest and returned to shore.

Next, the two chiefs took their throw nets. The Tinian chief cast his net along the
shore. When he pulled the net up, it was full of fish. Taga was worried. He did not know
Tinian and did not know where the good shore fishing spots were located. He finally
gave his net a great heave and threw it far beyond the reef into the deep water. He dove
into the open sea, pulled up his net, and found there were just as many fish in it as in the
net of the Tinian chief. Because Taga had thrown his net so much farther, he won this
contest.

After this, the two chiefs took their fish home to cook them. They needed some coco-
nuts for coconut cream to put on the fish. The Tinian chief started to climb a coconut
palm. ‘That is not the way,’’ said Taga. Taga grasped the coconut palm and shook it so
hard by the trunk that even the smallest coconuts fell off. The Tinian chief was shamed.
Twice he had been defeated.

‘‘We need now to husk and grate the coconuts,’’ said the Tinian chief. He husked
several coconuts, broke them open, and gave them to his wife to grate. ““That is not the
way,” said Taga. Whereupon Taga took a nut, tore off the husk, and crushed the nut in

90 MARIANAS PREHISTORY

his two hands, squeezing out the cream. The Tinian chief had been defeated again.
Three times he had been defeated. So Taga became the chief of Tinian.

Then Taga built himself a great house. The ruins of this house are still called the House
of Taga.

After the house was finished, Taga’s wife became pregnant. At this time, Taga de-
cided to go to Saipan, to fight with anyone who wished to oppose him. So he went to
Saipan with his wife and found a cave as shelter for her. But at night she gave birth to a
son. The cave was contaminated with birth fluid. Taga was afraid to fight, as he might
lose his vitality by being present at a childbirth, so he returned to Tinian.

When Taga’s son was five years old, Taga caught a beach crab, tied a string to it,
and gave it to his son to play with. The string broke and the crab ran into a hole under a
coconut tree. The child called, ‘‘Father, come here and push over the tree so that I can
get the crab.”

Taga refused, because the coconut was one he had planted, and it had just com-
menced bearing nuts. The boy became very angry. He pushed the tree back and forth,
and finally pushed it over. Then Taga became afraid. ‘“The boy is so strong he may kill
me when he gets older,’’ Taga thought. So that night when the boy was sleeping, Taga
killed him by strangling him.

Then Taga’s daughter became very afraid of her father, and also sorrowed over her
brother’s death. She fled into the Carolines—the mountains of Tinian.

The mother, too, sorrowed so that she became ill and died. Taga was filled with re-
morse. One day Saint Joseph came to Tinian to convert the people to Christianity.
Taga became a Christian and died on Tinian.

ARCHAEOLOGICAL WorRK AT THE HousE oF TAGA

The expedition’s archaeological work consisted of bushing and cleaning
the site as a public service, mapping and measuring the shafts and cap-
stones, and excavating a series of strati-tests. A ground plan of the shafts
and capstones is given in figure 37 and photographs of sections of the site
are shown in figures 38 and 39. Description of the latte follows:

Latte

Weathering—Both shafts and capstones show a certain amount of
weathering. The sides and ends of the shafts, as well as the sides of the cap-
stones, exhibit parallel striations caused by differential weathering. Also,
both shafts and capstones have tended to develop pockets of softer stone.

Material.—All shafts and capstones are coral limestone.

Form.—Shafts are moderately trapezoidal. Capstones are cup-shaped.

Constructional Features—The House of Taga was literally built on sand.
Shafts were sunk 2 or 3 feet into the ground, but without adequate founda-
tion. This has undoubtedly contributed to the collapse of the shafts.

NOTES ON INDIVIDUAL SHAFTS AND CAPSTONES
Shaft 1: Shaft fallen outward and cracked transversely about one-third
of the distance from the top. Capstone rests on side, with lower margin
buried in ground.
Shaft 2: Shaft fallen inward but it remains unbroken. Capstone rests
on side.

*polsquinu sauojsdeo pure syeys 407 ‘uRIUTT, ‘soyduot 3s9} Jo pue ese] jo asnopzy jo uefg ‘“/¢ ‘Oly

» v 1S3L ‘
ing
ee
> xX
q = 1
° u S S 9
3
y ‘
Q a isaL 2 »
Q isa
z+. - :
% ’
™
bs)
.y
md 8
l

Ol |
9 1S3L es
[| Vane

i:
yA

ging 4aeass
94as2U09

92 MARIANAS PREHISTORY

Fic. 38. Shafts 4 and 9, House of Taga, Tinian; looking southwest. Fragment of
capstone rests in place on top of shaft 9, shown at right.

Shaft 3: Shaft fallen outward. Capstone rests on side. Top of shaft buried
only 0.1 foot below the present surface, while lower margin of shaft was
found virtually on the surface; probably this shaft fell only recently.

Shaft 4: The only intact, standing shaft. Shaft is erect and capstone re-
mains in position on the shaft. However, a large piece of the capstone has
cracked off and has fallen inward, and another piece has fallen outward,
breaking into many fragments.

Shaft 5: Shaft fallen outward. Base is almost buried up to the top surface,
and this exposed upper surface shows much weathering. Capstone rests on
side, with lower margin buried in earth. Upper face of capstone pock-
marked by weathering. Relatively greater weathering of exposed upper

SURVEY AND EXCAVATIONS ON TINIAN 93

Fic. 39. Fallen shafts and capstones, House of Taga, Tinian; looking southwest.
Capstone of shaft 3 in foreground, shaft 1 with capstone in background.

surfaces of both shaft and capstone suggests that this shaft was one of the
first to fall.

Shaft 6: Shaft fallen outward, with base of shaft half buried. Capstone
rests on side, with upper surface showing considerable weathering.

Shaft 7: Shaft fallen inward, with base partially buried. Capstone rests
on side, with lower margin partially buried.

Shaft 8: Shaft fallen inward, with base partially buried. In falling, the
capstone struck shaft 5, shattering the capstone, although its form is
identifiable from the fragments.

Shaft 9: Shaft still standing but upper half has broken off, dropping the
capstone to the ground. Capstone rests upside down, the only one in this
position.

94 MARIANAS PREHISTORY

Shaft 10: Shaft base remains upright and in place, but upper section of
shaft has broken off, falling inward. Capstone was dropped between shafts
2 and 3 and rests on side.

Shaft 11: Shaft fallen outward, with base partially buried. Capstone
rests on side.

Shaft 12: Shaft fallen outward. Capstone rests on side.

Isolated Capstone: One peculiar capstone, not belonging to any of the
shafts, lies half buried next to the fallen shaft and capstone of Shaft 3.
This capstone is small (diameter 3.45 feet), and its cup-shaped surface has
been reworked to form three concentric steps. It may have been placed as
the base of a ladder into the house.

Dimensions: The intrinsic interest of the House of Taga makes it de-
sirable to give the dimensions of the shafts and capstones. There is con-
siderable variation among them. All dimensions are in feet.

Excavations

Inspection of the surface within the area of the Jatte itself revealed a suf-
ficient number of depressions to indicate that the site had been very con-
siderably subjected to previous excavation. Also, there were very few
sherds on the surface. Hornbostel noted that at the time of his visit (1924)
he collected in a 10 X 10-foot area on the surface 39 pounds of pottery
fragments. Their scarcity today indicates the thoroughness with which
visitors have picked up surface artifacts.

In view of the surface disturbance in the immediate area of the latte, a
section on the east side of the site was selected for four test trenches. The
site was bounded on the east by a concrete street curb and three trenches
were located in the old Japanese street. The Taga shafts are on a low rise
approximately two feet high. The Japanese street was cut through this
mound but it was never surfaced. It was not used by American forces
after the invasion and today leads nowhere. Except for the earth removed
by the Japanese, the street was completely undisturbed, for no sewer or
power lines had been laid below it. The fourth trench was located on the
other side of the curb, just east of the latte, in an undisturbed location.
These test trenches are shown in plan (fig. 37) and in section (fig. 40);
excavation details are given below.

Test Trench A: In constructing the street, the Japanese removed approxi-
mately one foot of over-burden, but the subsurface area below was undis-
turbed. The trench was excavated to a depth of 4.0 feet. No pits, ash
deposits, or occupation levels were found, but sherds occurred to a depth
of 3.6 feet. The excavated material was all beach sand, and cave-ins of the
trench prevented excavating to a greater depth.

8° el
cL

ona
rw

Dey, OF OS iS
o£ TES Sor
BAL Saye Pas
9 g v &
SANOLSdVO

Ait Le s‘¢
o¢ 05% aera C¢
» oF 82S c+
Be Cue Gat

8° cl SO VE «2 Tl Web
9 g v &;
SLAIVHS

‘g0Rjans punois sAoqe sjd0fo1g ,

8b Gb octets ssouyory J,
0O'+ Ob tit 90Ry Jamo]
G8iL Zhi ee ear ey 20RJ sroddy,
I9}oUIe IC]
A L
re Gg curtis won0g
GIGiC p'¢ Shieh & S16 Owe) Se. 6) 81 wR doy,
ssouyory J,
0's Cg ttt wonog
6u¢ LAG aioe. we wali Br 'e8) HAY doy,
GIPIM
Zz CI 6 Zl + €, 0 a e166. 8 we wee yysuo'T
cA L

96 MARIANAS PREHISTORY

Test Trench B: In this trench, approximately two feet of over-burden had
been removed during the construction of the street. The material removed
was all beach sand. No pits, ash deposits, or occupation levels were found.
Trench was excavated to a depth of 4.0 feet. Lowest sherd was found at
depth of 2.3 feet.

Test Trench C: Approximately one foot of over-burden had been removed
during street construction. Soil composition same as in Trenches A and B.
At a depth of 2.5 feet a few small patches of dark gray sand were found,
but there were no other signs of pits or occupation levels. Trench was
excavated to depth of 4.0 feet. Lowest sherd was found at a depth of 3.5
feet

Test Trench D: No over-burden removed previous to excavations. This
allowed the obtaining of a soil profile of an undisturbed area (fig. 40) and
made it possible to determine the pottery association of the Taga Jatte.
Trench was excavated to a depth of 6.0 feet. Lowest sherd was found at a
depth of 5.3 feet.

Burials

Just below the surface in Trench B, a few fragments of human pelvis and
of the bones of the lower extremities were uncovered. Apparently, most of
the burial was removed when the street was put in, during the period of
Japanese administration. No other burials were found in the trenches.

GENERAL DISCUSSION

Together with the As Nieves Site on Rota, where the shafts and cap-
stones are even larger than those at Taga but are still located at the quarry
and not at their ultimate destination, wherever it may have been, the
house of Taga represents the climax of /atte building in the Marianas. A
question arises as to what sort of structure was placed on top of the stone
latte. Was a roof placed directly on the capstones and shafts, or was a floor
secured to the capstones with a wood structure above it? The top surface
of the capstones was about sixteen feet above the ground, which would
have resulted in a curious-looking structure, if a house were perched atop
the latte. The answer to this question must remain one of speculation.

Another question is the manner of erecting the latte, as the weight of
shafts and capstones is so great that some mechanical means must have
been used. The simplest explanation is that the latte were set in place
through the use of earth ramps. Here again any explanation is largely
speculation.

The amount of surface pottery reported by Hornbostel, as well as that
recovered from the upper level of Test Trench D, indicates that a good
deal of ordinary daily living went on around the House of Taga. Whether

SECTION Xx!

SECTION yyt

SECTION zz!

Test B

ate ZePetefee%e®.

> 7
ahr te

raya veva

10 feet

trenches, House of Taga, Tinian.

98 MARIANAS PREHISTORY

it had a specialized use cannot be determined. The original form of the
Taga Site consisted of a series of latte structures strung out along the beach,
with the largest—the House of Taga—near the center. This arrangement
paralleled that found at the Blue Site. It is known that the Chamorros at
the time of contact had special men’s houses, and the House of Taga may
have been one. The Chamorros also maintained sharp class distinctions,
and the House of Taga might have been the residence of a chief. In any
case, I believe it most likely that the Taga /atte reflected the social structure
of the community just as do the men’s houses of parts of western Micro-
nesia.

The test trenches demonstrated that the House of Taga was only the
last manifestation of a considerable period of occupation. With sherds
found at a depth of over five feet below the surface, the site must have been
occupied for a good many years. This matter is considered in greater detail
in the conclusion of this report.

THE TAGA QUARRY

On the shore, approximately 4,000 feet south of the House of Taga, is
situated the quarry from which Taga’s giant shafts and capstones came.
It is located on a low, limestone bench that terminates along its western
side in a cliff that falls abruptly to the sea. A map of the quarry is shown
(fig. 41). This is the only quarry found on either Tinian or Saipan.

The quarry forms a series of distinctive patterns of shaft and capstone
holes. Along the east edge the pattern becomes confused and less well de-
lineated. There are numerous overlaps of quarry holes, some of which are
difficult to explain, unless a shaft or capstone broke in the process of
quarrying.

The quarry is separated into two parts by a small cove that cuts into the
limestone bench. At the cove, several great pieces of limestone have broken
off the edge and fallen into the sea. In at least one case, a large section of
limestone has fallen after shafts have been quarried from it, for the quarry
holes are terminated by the cliff edge (see fig. 41).

The quarry holes themselves are not completely symmetrical. Appar-
ently a small trench was cut around the capstone or shaft and this trench
was deepened until the stone was undercut and freed from its matrix. In at
least three cases, shallow trenches run into the quarry holes. Their use is
unknown, unless it was to facilitate the collection of water in the trench cut
around the shaft or capstone. Judging from the curvatures of the quarry
holes, capstones were taken out right side up, in the same position as when
placed on the shafts at the house site. In a few places, tool marks can be
seen on the sides of the quarry holes. The marks appear as vertical, parallel

‘uelury, ‘Atien() Vesey jo ueg

“Ty “Old

w
VIS INIdd/TIHd

Fic. 42. Taga Quarry, Tinian. Upper: quarry hole of capstone. Lower: quarry
hole of shaft. Level rod is 6.5 feet long.

100

SURVEY AND EXCAVATIONS ON TINIAN 101

grooves and were probably made by adzes or chisels. Also there are a few
grooves to be found in the surface of the limestone, running into the edge of
quarry holes. These grooves are from 1.5 to 4 feet long and from 0.05 to
0.15 feet wide and deep.

It is evident from the map that this quarry was used not only for the
House of Taga shafts and capstones, but for smaller ones as well. How the
largest shafts and capstones were transported is a matter of conjecture.
There is no evidence that the ancient Chamorros knew and used the wheel.

IV. Archaeological Survey of Rota

Through the co-operation of the United States Navy, a brief archaeo-
logical survey was made of Rota in June 23-27, 1950. The coastal area was
surveyed along the north coast from Sosanlagh Bay to beyond Muchon
Point, and along the south coast from Sosanjaya Bay to Mariiru Point. In
addition, an examination was made of the very interesting interior As
Nieves Site.

Rota abounds in coastal /atte sites, with the greatest concentration along
the north coast, particularly in the Muchon Point area (fig. 43). At the
latter location there are literally dozens of latte, including the only 14-shaft
one (over 72 feet long) that I observed. Some are in clusters and others
more widely spaced. There has been considerable destruction of Rota Jatte
through the years, particularly as the Japanese during World War II con-
structed a coastal trench system, which disturbed numerous sites. Never-
theless, Rota retains more Jatte sites intact than either Guam, Tinian, or
Saipan. The cultural deposits at these sites, however, do not seem to be
deep and I was unable to locate sizeable refuse deposits.

The logical spot for the first migrants to have settled on Rota is at the
site of the present settlement. This area has been much disturbed, as it was
the seat of the Japanese sugar mill and the adjacent town. No Jatte were
found here, though surface sherds are common. A considerable part of the
area consists of limestone sands, and a series of test trenches might reveal an
underlying pottery seriation, with Marianas Red sherds at the lower levels
as at the Chalan Piao Site on Saipan.

The interior As Nieves site rivals the House of Taga on Tinian in im-
pressiveness. At As Nieves, a series of nine giant shafts and seven capstones
destined for a structure larger than the House of Taga remain at the spot
where they were quarried. With one exception, the shafts and capstones
are still in the quarry holes, with trenches cut around each shaft and cap-
stone. The prehistoric Chamorros of Rota were bent on erecting a giant
latte structure, but some unknown event caused them to abandon their
work after it was well on the way to completion. Just where they proposed
to erect the shafts and capstones is unknown.

The As Nieves site was first carefully mapped and described by Horn-
bostel in 1925. His plan and elevations, taken from his notes, are in-

102

*ALIEN() SPAIN] SY PUB SOzIS 27777 Jo seoie [edioulid Sutmoys ‘e}oy jo depy “Cp ‘OL

leSbl

,9 —,9
‘Ld NUILYVAN
AVG VAVPNVSOS
8 of 8
jet) . bbl
fae
A’; HOVINVSOS

—i6

sit

oct

aoot
ey; prover —i01

PI ‘Ld OHOWLVL f

rAl ——
: ‘Ld NOHONW rr

\
781 9 vl ,cleGvl /0l A)

103

‘yydop ut 3995 Zp 0} Q'Z Woy soyozTp 19yI0 ‘daap 100F CE") Ajuo Z yeys punose youd
‘uI9y) puNose syURq SWIC SOyOIP Wo [FY “PUOISOUTTT 942 OFUT PoBABOXS SYP Aq papunosms ‘aoejd ut ureures souoysdeo pue
syeys royj0 [Ty ‘efoy Arrenb szt jo yno pari] Useq sey ¢ suojsder) *([a}soquIOFy Joye) eIOY “SOASIN' SY 7% Aizenb jo uejg ‘py ‘Oy

a My, Wer tine yy
> Wy, 7
ie . 3 4
: ” s
2 % ‘
—t T t ene”) ci 2 =
1995 OF 0 ot 5 & lore
aTvos e: 4 <

= FA Jute a a
cata ve ag
= ae
= “ 5
: ° g
i youtP a
z

€ La4vHs 1 LAvHs z
: preeuwer 1 z
: =
id +
Fair ayScteneTONAS action wEASE | AN LEI UCU Baer

uw A?
wt Me
yw %
Ps ve We
E: a ot “Un,
% on te.

N 2 RS) 4
z 4
=
2 a 2
¥ = =

= ie z
my <€
Fd
\ > a4 4 é
ww = § fe 3
awe Seo, < : < . %, =
1 a = fe ar
wa > s v
a = : < “a >
> S z tn ois
Cas = 1 Tyga
=: E é
Rk $ a
<2 = =
Se x = we Me
3 MAW oe 4,
Siete
Ss
= % 4 ~
bY on nae Ai aw
“m = neers
AWN ayy we i)

104

~<
~<
=
Oo
—
o
lu
”

106 MARIANAS PREHISTORY

cluded in figures 44 and 45. His observations of As Nieves provided Horn-
bostel with a clue to the method used in quarrying the giant shafts and
capstones. In examining the banks of soil thrown out of the quarry holes,
he noted that these banks contained considerable lime and fired rock par-
ticles. Hornbostel thereupon concluded that the prehistoric people at As
Nieves built fires in the ditches around the shafts and capstones. These
fires reduced the coral limestone to lime, or at least to friable material,
which could then be removed with the basalt adzes of the ancient Chamor-
ros. Hornbostel’s explanation is entirely plausible, particularly when it is
remembered that the prehistoric inhabitants of the Marianas were betel-
nut chewers, and hence probably familiar with this method of obtaining
lime from coral limestone to combine with betel nut. If this is true, it is an
interesting example of the transfer of a trait associated with chewing a
narcotic to the quarrying of stone.

Vc Pottery

A principal characteristic of archaeological sites in the Marianas is the
abundance of pottery. Over 20,000 sherds were recovered during the
course of archaeological work on Saipan and Tinian. Of all artifacts, pot-
tery is the principal time-indicator of culture change in these islands.

The absence of whole vessels is remarkable. Despite the abundance of
sherds, only one restorable vessel was found during the entire period of
field work. As a result, pottery types have had to be established on the
basis of sherds alone.

In 1698, Saipan and Tinian were conquered by the Spanish after many
years of intermittent warfare. The Chamorro inhabitants of both islands
were forced by their conquerors to move to Guam. Saipan was resettled in
the nineteenth century by a colony of Carolinian islanders from the central
Carolines and later by Chamorros from Guam. The Carolinians were
from islands where pottery was not made, and the Chamorros who re-
turned to Saipan had been under western influence so long that pottery-
making had disappeared among them. During the eighteenth and nine-
teenth centuries, Tinian was stocked with cattle by the Spanish, who pe-
riodically sent parties to the island, where they hunted the cattle running
wild on the island and returned to Guam with the meat, but no town was
established on the island and no more than a very small handful of persons
was ever on the island at one time. It is quite definite that the pottery
found at archaeological sites on Saipan and Tinian predates the abandon-
ment of these islands at the time of final Spanish conquest in 1698.

METHODS OF EXCAVATION AND ANALYSIS

In the excavation of archaeological sites on Saipan and Tinian, pottery
and other artifacts were removed in 0.5 foot levels and by squares. Both
distinctions were preserved in the analysis of the sherds. However, the lack
of any statistically significant horizontal differences in pottery type dis-
tribution led to the grouping of squares in the presentation that follows.
Levels proved to be significant; horizontal distribution did not.

All sherds were washed and given preliminary examination in the field.
A considerable number of Marianas Plain body sherds were typed, counted

107

108 MARIANAS PREHISTORY

and discarded in the field. All rim sherds and all sherds of types other than
Marianas Plain were shipped to the Museum for further analysis.

POTTERY “LYPES

The pottery obtained during the course of excavations on Saipan and
Tinian has been classified primarily from the point of view of usefulness in
defining temporal and local differences. Another archaeologist might have
imposed a classification of greater refinement on the sherd material if his
objective was simply to describe and classify on the basis of small differ-
ences in paste quality and similar characteristics. ‘This applies particularly
to the largest category of sherd material, included in the type here given
the designation Marianas Plain.

Marianas Plain is the most abundant type in the Marianas. Out of it
probably grew Marianas Fine-Line Incised, Marianas Cord-Marked, and
Marianas Trailed, which are differentiated from Marianas Plain on the
basis of surface decoration. These decorated types are numerically insig-
nificant on Saipan and Tinian, but I suspect they may be more significant
on Guam. This supposition is based on only a casual acquaintance with
Guam pottery, but there may be important local differences between
Tinian and Saipan on one hand and Guam on the other (Thompson, 1932,
p. 28).

The fullest previous description of prehistoric Marianas pottery is in the
report by Thompson (1932) on the Hornbostel collection now in the Bishop
Museum. Of the 1,559 sherds in this collection, Thompson assigned 1,546
to her Type 1; 6 sherds of a type not found in my own field work to her
Type 2; 2 red-slipped sherds to her Type 3; and 1 burial urn to her Type
4. Four indeterminate sherds were not classified.

Of the pottery types described in this report, Marianas Plain, Marianas
Fine-Line Incised, Marianas Cord-Marked, and Marianas Trailed all fall
in Thompson’s ‘“‘Type 1.”’ Also, in accordance with my own classification, I
would assign Thompson’s ‘Type 4’ (burial urn) to Marianas Plain.
Thompson’s ‘‘Type 3” (red slipped ware) coincides with Marianas Red.
One of Thompson’s ‘‘Type 1” sherds I would assign to the lime-filled,
impressed trade ware described hereafter.

The classification of pottery types that follows represents partly a refine-
ment of Thompson’s ‘Type 1’’—a category in which she placed all but a
few of the sherds from the Hornbostel collection. It should be noted that
Thompson was forced to devise her sherd classification with only very
minimal stratigraphic information.

POTTERY 109

MARIANAS PLAIN

General.—Of all the pottery types found, Marianas Plain is by far the
most common. It is particularly associated with J/atte sites, where it occurs
in great abundance. The following description is based on megascopic
observation of over 19,000 sherds. Typical sherds are shown in figure 46.

Fic. 46. Marianas Plain sherds. Hole bored in upper left sherd for suspension
of pot.

Provenience.—Objan (5,648 sherds); Laulau House A (1,976 sherds);
Laulau Rock Shelter (1,103 sherds); Oleai (296 sherds); Chalan Kija (413
sherds); Chalan Piao (586 sherds); Taga (1,275 sherds); Blue I (6,887
sherds); Saipan Survey (68 sherds); Tinian Survey (862 sherds); Rota
Survey (100 sherds). Total 19,214.

Paste-—There is great variation in the technical quality of this pottery
type. Sherds vary from poorly fired and friable examples to well-fired, hard

110 MARIANAS PREHISTORY

sherds with a well-compacted paste. Color of core varies from brick-red
through brown to dark gray. Sherds are grit-tempered, except for a few
examples that also contain minute white inclusions (ground coral?). Ves-
sels were probably modeled. One sherd shows interior finger-modeling
impressions. No evidence of coiling. Hardness of core varies from 2.5-4.5
(Moh’s scale).

Surface.—Undecorated except for two examples of small lugs and a num-
ber of examples of finger-impressed or incised decoration applied to the
lips of rim sherds (see section on Form below). Predominant color of surface
is reddish brown, though color may vary to a brownish gray. In the case of
poorly fired sherds, surfaces tend to be eroded and crumbly. Exterior sur-
faces smoothed. Interior surfaces are sometimes smoothed for about 5 cm.
below the lip. Sherds are unslipped, with the possible exception of some
sherds that exhibit a very thin grayish film that may be merely a calcareous ~
deposit. The one restorable vessel exhibited the marks of pandanus woven
matting on the bottom surface. One other sherd also showed a mat impres-
sion. As these mat impressions may be merely the result of modeling the pot
on a matting surface, I did not feel justified in establishing a separate pot-
tery type to accommodate this form of surface treatment. Some sherds have
a deposit of carbonized material adhering to the inner surface, indicating
that these sherds came from cooking pots. Surface hardness 2.5—5.5 (Moh’s
scale).

Thickness—A sample (which I believe to be representative) of 2,200
body sherds was measured for thickness. Measurements are given in the
frequency graph below (fig. 47). The relatively large number of sherds
measuring more than 12 mm. in thickness is one indication that this pottery
type includes an appreciable number of large, heavy vessels.

Form.—Despite the large number of sherds found, only one restorable
vessel, from the lower levels of the Taga Site and probably atypical in
form, was recovered. This vessel is shown in figure 48. Marianas Plain ves-
sels seem to have run to large sizes. Sherd curvatures indicate that pots up
to 50 cm. in diameter were made. One Marianas Plain bowl from Guam
was illustrated by Thompson (1932, pl. 5). This vessel is distinctive in that
four small lizards are modeled in low relief on the exterior surface just be-
low the rim. This type of relief decoration was not found on any Saipan or
Tinian sherds. One large sherd from a flat plate (22 cm. in diameter) was
found at Blue I site. Four rim sherds from Blue I showed holes for pot
suspension,

All Marianas Plain rim sherds were carefully analyzed to determine dif-
ferences in frequency among the various levels of a single site, as well as
between sites. Rim types are as follows (see fig. 49):

POTTERY 111

1000 +

800 F

700 F

600 F

tLe
“TW...

4 8 12 16 20 24 28 32° mm,

Fic. 47. Graph showing thickness of Marianas Plain body sherds, based on
sample of 2,200 sherds.

Rim Type A.—In cross section, interior and exterior surfaces are either
parallel or taper slightly toward the lip. The upper surface of the lip is
usually rounded but occasionally is flattened. This type seems to be associ-
ated with straight-sided vessels, except for a very few sherds which exhibit a
constricted neck and an everted lip.

Rim Type B.—In cross section, there is a pronounced enlargement at the
lip. The upper surface is predominantly rounded but occasionally is flat-
tened. Also, a very few sherds show a markedly everted lip.

Rim Type C.—Like Type A or B, except that the upper surface of the lip
or the exterior upper surface of the rim exhibits a regular pattern of finger
impressions. This type is low in frequency.

la MARIANAS PREHISTORY

Fic. 48. Marianas Plain bowl, top view. ‘This pottery vessel is similar in form to
carved wood bowls from the Carolines and some other parts of Oceania; this similarity
suggests that the vessel is derived from a carved wood prototype.

Rim Type D.—Like Type A or B, except that the upper surface of the lip
exhibits a regular pattern of transverse incised lines. The upper surface of
the lip is flat. Low in frequency.

Both Type A and Type B can be subdivided according to whether the
upper surface of the lip is rounded or flattened. This subdivision was first
made in the analysis of rims, but as the resulting division showed no statisti-
cal significance for purposes of seriation, the subdivision was not made in
the final analysis. The results of the seriating of rim types are given in a
following section.

A few rim sherds were so fragmentary they could not be classified. A few
others, of exotic form, were also not classified.

MARIANAS FINE-LINE INCISED

Paste, thickness, color, and form are the same as those of Marianas
Plain. Distinctiveness of pottery type is based on surface decoration. This

\b}) INC

J) (ac

NIPVE
Cty

Fic. 49. Marianas Plain rim sections. Upper two rows: Rim Type A. Bottom
two rows: Rim Type B. All sherd sections shown with sherd exterior to right. One
half natural size.

013

114 MARIANAS PREHISTORY

consists of a pattern of narrow incisions on the exterior surface of rim and
body. Incisions on Tinian and Rota specimens are in the form of a cross-
hatch pattern (fig. 50), though Thompson reports other types of fine-line
incising (Thompson, 1932, pp. 27-28), presumably from Guam. Though

Fic. 50. Upper row: Marianas Plain rim sherds (Type C) with finger impressions
around lip. Lower row: Marianas Fine-Line Incised sherds.

the pottery type is numerically insignificant on the basis of the Museum’s
work at Tinian and Saipan, the type has been set up here because there
may be marked differences in its frequency between Guam on one hand
and Saipan and Tinian on the other.

Provenience.—Blue I (3 sherds); Rota Survey (4 sherds).

MARIANAS CORD-MARKED

This pottery type is the same in paste and thickness as Marianas Plain.
The forms were probably also the same, except that the few sherds found
of this type do not admit of a definite answer. The pottery type is distin-
guished by the manner of surface decoration (fig. 51). This decoration is
either regular (a series of parallel lines) or random (sets of parallel lines,
with different sets overlapping and running in different directions). The

POTTERY 115

difference is best seen in figure 51. On the few rim sherds found, parallel
lines of the regular form of decoration run vertically downward from the
lip. Impressions are from 1 to 2 mm. wide.

The decoration was probably impressed into the surface of the vessel by
a cord-wrapped paddle. Solheim (1952) has recently reviewed the evidence

Fic. 51. Marianas Cord-Marked sherds. Middle sherd in left column random-
marked; other sherds regular-marked. Note how impressed lines proceed vertically
downward from lip of vessel, as shown in two upper sherds.

for paddle decoration of pottery in Oceania and Malaysia and in naming
this type I have followed his conclusions. At the same time, I was unable to
determine with certainty whether all sherds assigned to this type exhibited
cord-impressions, or whether some may have been combed. Yet from the
widespread nature of paddle-decorated pottery in Malaysia and its exten-

116 MARIANAS PREHISTORY

sion into the Marianas, I believe the assignment of these questionable
sherds to the cord-marked category is valid.

Provenience.—Objan (7 sherds); Taga (4 sherds); Blue I (24 sherds);
Tinian Survey (1 sherd); Rota Survey (18 sherds). Of this number, only 6
sherds, all from Blue I, exhibited the random form of cord-marking.

Fic. 52. Marianas Trailed sherds.

MARIANAS TRAILED

This pottery type is the same in paste, thickness, color, and probably in
form as Marianas Plain. Distinctiveness of pottery type is based essentially
on surface decoration. This is formed by trailed impressions, from 4-10
mm. in width, on the exterior surface of the vessel. In the case of the few
rim sherds found, the trailing proceeds vertically downward from the lip
(fig. 52). The trailed impressions are parallel. They were probably exe-
cuted with the fingers, though an instrument may have been used.

POTTERY i We

Provenience:—Objan (1 sherd); Blue I (3 sherds); Tinian Survey (2
sherds).

MARIANAS RED

General.—The following description is based on megascopic observation
of 485 sherds. No whole vessels of this type were found and inferences re-

Fic. 53. Marianas Red sherds. Rim of lower left sherd has broken off, leaving a
fracture that suggests use of coiling technique; the sherd also exhibits erosion of red slip.

garding vessel forms are based on analysis of sherds alone (fig. 53). Techni-
cally, this is a well-made pottery type, superior to Marianas Plain.

Provenience—Chalan Piao (397 sherds); Laulau Rock Shelter (81
sherds); Oleai (6 sherds); Blue I (1 sherd).

Paste.—Even-textured and well compacted, though some Chalan Piao
sherds are slightly laminated and somewhat friable. In the range of sherds
examined, the color of the core varies from dark gray to gray red. Paste
contains numerous white inclusions (finely ground coral?) probably used
as temper. From the sherds it is very difficult to tell whether coiling was
used, though some rim sherds suggest this technique. Hardness of core
2.5—3.5 (Moh’s scale).

118 MARIANAS PREHISTORY

oa
Oo
'

ea]
=)
1

YY:
YY

FREQUENCY
AN
°
1

30-

SHERD

SSS. ISS 1 1 iN [mf
Zz 4 6 Ss 10h 12. 47 lone Bea. B20 a22 245 264) mms

Fic. 54. Graph showing thickness of Marianas Red body sherds, based on
sample of 165 sherds.

Surface.—Predominantly red, though some sherds vary to red-brown, or
red-gray. Thin red slip applied to exterior surface and in a few cases to
interior surface as well. Slipped surfaces smooth and often glossy. Some
Chalan Piao sherds show considerable erosion of surfaces, and in many
sherds only a part of the slip remains. Occasionally it is possible to flake
off the red slip with the point of a knife. Except for two sherds with lip
impressions, no decoration was applied to surfaces. Hardness of surface 3.5
(Moh’s scale).

T hickness—Predominantly a thin-walled ware, in contrast to Marianas
Plain. Measurements of the thickness of 165 body sherds, given above (fig.
54), confirm this gross impression, which was most striking at the time of
field excavation.

Form.—Analysis of rim, shoulder, and bottom sherds indicates that ves-
sels were predominantly small. Both bowls and jars are represented in the
collection. Range of rim sections is shown in the accompanying figure. The
most common type has an everted lip. A few body sherds exhibit a well-
defined shoulder. Bottoms of vessels seem to have been predominantly
rounded, but four sherds gave definite evidence of a flat base.

Variant Sherds.—In addition to the 485 sherds on which the above de-
scription is based, we excavated an additional 164 sherds which varied
from the Marianas Red sherds in surface color, though otherwise they were

POTTERY 119

AIC EEC
ir)

yd»

a fie ge

Fic. 55. Marianas Red sherd sections. Upper two rows: rim sherds. Third row:
shoulder sherds. Bottom row: base sherds. One half natural size.

technically similar. Some of these sherds verged toward a buff, but the
majority toward a dark gray. The provenience of these sherds is as follows:
Oleai (1); Laulau Rock Shelter (11); Chalan Piao (152). The distribution
in depth of these variant sherds showed a clear association with Marianas
Red. Additional excavation may indicate the desirability of setting up a
separate pottery type for the dark gray sherds, but I have purposely re-
frained from doing so in this report, as further extension of the formal clas-

120 MARIANAS PREHISTORY

sification did not seem warranted. Additional excavations in the Marianas
may reveal that the dark gray sherds on one hand and the Marianas Red
sherds on the other are associated with culturally established, consciously
intentional differences in firing technique (oxidizing vs. reducing) rather
than accidental differences. In this case, the dark gray sherds should be
established as a separate type.

LIME-FILLED, IMPRESSED TRADE WARE

General.—This is a highly distinctive pottery type, set off from the others
by the manner of decoration and the technique of manufacture. Only
twelve sherds of this type were found, indicating a probable trade ware.
For this reason, no locality designator has been used. However, the area of
origin is unknown. The following description is based on megascopic ob-
servation of twelve sherds.

Provenience.-—Laulau Rock Shelter (11 sherds, found in a small horizon-
tal area, at a depth of 2.4 to 2.8 feet); Chalan Piao (1 sherd, found within
top 0.5 foot).

Paste—Somewhat friable, slightly laminated, and containing numerous
minute white inclusions (ground coral?) as well as particles of grit. Color of
core dark gray. Fractures indicate that this pottery was made by coiling
technique, and that the sherds broke off at the coils, leaving smooth-edged
horizontal breaks and rough-edged vertical breaks. These sherds provide a
clear example of coiling in prehistoric times in the Marianas. Hardness of
core 3.5 (Moh’s scale).

Surface-—Color of surfaces dark gray. However, interior surface of one
sherd has a thin red slip, and there are faint indications of a thin reddish-
brown slip on the interiors of six other sherds. Exterior surfaces smoothed,
but neither polished nor slipped. Hardness of surface 3.5 (Moh’s scale).

Thickness.—Range of variation of thickness of body sherds 9-14 mm.

Decoration.—Eleven sherds were decorated on the exterior with im-
pressed circles, circles with interior dots, and bands of impressed, opposed
chevrons or of parallel vertical lines, or combinations of these. One rim
sherd was undecorated on the exterior surface but possessed an incised, zig-
zag line 4 mm. below the rim on the interior surface. These impressions and
incisions had been filled with lime, which made the decoration stand out
strikingly from the dark gray of the surface. Figure 56 illustrates the distinc-
tive nature of the decorative treatment.

Form.—Seven rim sherds were obtained, all having a similar cross sec-
tion. In the absence of whole vessels of this type, it is possible to state only
that the rim sherds suggest open mouth jars, or bowls. The question arises
whether the lower parts of vessels of this type were undecorated and hence
have possibly been assigned to Marianas Plain in the sherd analysis. Every

Fic. 56. Lime-filled, impressed sherds.

Fic. 57. Rim sections of lime-filled, impressed sherds. One half natural size.

121

122 MARIANAS PREHISTORY

SITE POTTERY TYPE FREQUENCY TOTAL
SHERDS
2000 3000 4000 $000 6000 7000
MAR. PLAIN 5,648
OBJAN MAR. CORD-MARK. 7
MAR. TRAILED f
MAR. PLAIN 6,887
BLUE I MAR. FINE-LINE INC. 3
MAR. CORD-MARK. 24
MAR. TRAILED 3
MAR. RED 1
MAR. PLAIN 1,275
TAGA MAR. CORD-MARK. 4
MAR. PLAIN Ene 1,976
LAULAU
HOUSE A
MAR. PLAIN Bag ; 413
CHALAN
KIJA
MAR. PLAIN 296
OLEAI MAR. RED 6
MAR. RED(VAR.) i
MAR. PLAIN 1,103
LAULAU MAR. RED 8!
ROCK MAR. RED (VAR,) "
SHELTER LIME-FILLED |MPR.
MAR. PLAIN 586
CHALAN MAR. RED 397
PIAO MAR. RED (VAR.) 152
LIME-FILLED IMP. fl

Fic. 58. Graph showing gross differences among excavated sites in relative fre-
quency of pottery types.

effort was made to examine all undecorated sherds with this point in mind,
but no undecorated sherds matching the technical characteristics of the
decorated sherds were found.

LOCAL DIFFERENCES AND TEMPORAL RELATIONSHIPS
REVEALED BY POTTERY

Gross DIFFERENCES AMONG EXCAVATED SITES

The accompanying bar graph (fig. 58) illustrates the gross differences
among excavated sites, when pottery types for each site are shown without
regard to levels of excavation. For all sites, Marianas Plain is numerically
the greatest. For latte sites, this numerical dominance is overwhelming, to
the point that at Laulau House A and at Chalan Kija no other pottery type
was found. It is clear that latte sites are characterized by a very high per-
centage of Marianas Plain.

Of non-latte sites, Laulau Rock Shelter and Chalan Piao are particularly
interesting because they contained an appreciable number of sherds of a
second major type, Marianas Red. The gross differences among sites sug-
gest that the key to temporal relationships among pottery types is to be

POTTERY 123

PERCENTAGE DIFFERENCES IN POTTERY TYPES

CHALAN PIAO

MARIANAS PLAIN

MARIANAS RED

MAR. REO~-VARIANT

LIME- FILLED

SURFACE

Ne i

LEVEL 1 QA AAA A
2 SSW AY
3 IY MAY
4 “‘\* ———
5 N \)
: \
7
8 N A \
; MAA
10 NAAN

N 10%

SCALE

LAULAU ROCK SHELTER

MARIANAS PLAIN

Q\\WWwv

MARIANAS RED MAR.RED-VARIANT

SURFACE
LEVEL 1

AuUhun

SCALE 10 %

Fic. 59. Graph showing percentage differences in pottery types. Upper: Chalan

Piao. Lower: Laulau Rock Shelter.

found in the relation of Marianas Plain to Marianas Red. This is precisely
the case.

TEMPORAL RELATIONSHIPS AMONG POTTERY TYPES

At two sites—Laulau Rock Shelter and Chalan Piao—a clear temporal
relationship was established between Marianas Red and Marianas Plain
(see fig. 59). At Laulau Rock Shelter, although Marianas Red was not
abundant, it occurred in the lower levels, phasing out toward the top levels.
Marianas Plain, though present in all levels, showed a lower percentage of
occurrence in the lower levels but was completely dominant in the upper
levels.

LIME—FILLED

124 MARIANAS PREHISTORY

The seriation data, therefore, lead to the following conclusion. Marianas
Plain, though probably always present in the Marianas, did not become
the dominant pottery type until late times. Marianas Red was either domi-
nant in early times or at Icast as important as Marianas Plain, but it faded
out with the passage of time. Marianas Plain is characteristically late (post
A.D. 500?); Marianas Red is characteristically early. It was not present in
historic times.

The occurrence of six sherds of Marianas Red in Levels 1 and 2 at Oleai
suggests that the site is early but permits no definite conclusion. The finding
of one Marianas Red sherd in Level 7 at Blue I is also inconclusive. Blue I
is dated at A.p. 854 + 145 and is a latte site. The single Marianas Red sherd
was found near the bottom of a refuse deposit and may well have been
originally a surface sherd deposited on the old ground level prior to the
time Blue I was occupied by Jatte-builders.

Returning to Laulau Rock Shelter, the occurrence of lime-filled im-
pressed ware in the lower levels only is also important. Presumably this is a
trade ware that was introduced into the Marianas relatively early and
never became established. The one sherd of this type found in Level 1
at Chalan Piao suggests that the site is early, but any conclusion based on
the presence of a single sherd is necessarily open to serious question.

Marianas Fine-Line Incised, Marianas Cord-Marked, and Marianas
Trailed seem to be types derived out of Marianas Plain. However, at
Laulau House A and at Chalan Kija only Marianas Plain was found. The
presence of iron artifacts at Laulau House A indicates that it was probably
historic and suggests the possibility that these derived types had passed out
of favor by historic times on Saipan. Yet Marianas Cord-Marked was
found in the upper levels at Objan, also undoubtedly a late site, as it is
described in Spanish seventeenth century accounts; so the evidence is not
conclusive. However, Marianas Fine-Line Incised, Marianas Cord-
Marked, and Marianas Trailed were nowhere found to be contemporane-
ous with Marianas Red.

CHANGE THROUGH TIME EXPRESSED BY MARIANAS PLAIN

With the temporal relationship between Marianas Plain and Marianas
Red demonstrated, the next question is whether within the Marianas Plain
category above, any change can be discerned. If such change were present,
it would provide another means of determining the temporal relationships
of the sites excavated, particularly those at which no Marianas Red was
found.

The gross differences in relative frequency of Marianas Plain rim types
among the sites excavated is shown in the bar graph (fig. 60). These data
are then translated into percentage differences (fig. 61).

SITE RIM FREQUENCY SHERDS
TYPE 200 300
A 34
LAULAU 8 19
HOUSE A c °
) °
uU 14
A 86
BLUE I 8 J 372
ic 28
D 7
uv 6
A { iat
OBJAN 8 =| 216
c 34
) 3
u 25
A 70
CHALAN 8 35
KIVA c °
D °
uv °
A | 77
OLEAI 8 | chy
c 2
) °
u !
A | 110
TAGA 8 ad
c | I
D | '
uv '
a 89
LAULAU 8 15
ROCK c ie
SHELTER ° 3
u °
A 40
CHALAN 8 2
PIAO c :
Do t
u 4

Fic. 60. Graph showing gross differences among excavated sites in relative fre-
quency of Marianas Plain rim types.

SITE TYPE 6 A ae Uh Gd 1)

LAULAU HOUSE A

BLUE I AI N |
OBJAN AI

CHALAN KIJA

OLEAI q

TAGA

LAULAU ROCK S.
CHALAN PIAO

SCALE 10%

Fic. 61. Graph showing percentage differences among excavated sites in relative
frequency of Marianas Plain rim types.

125

126 MARIANAS PREHISTORY

From the accompanying graphs it is clear that the principal point of
interest centers on the difference in frequency between Type A rim sherds
and those of Type B. On the basis of radiocarbon dates and pottery types—
principally Marianas Plain and Marianas Red—Chalan Piao and Laulau

MARIANAS PLAIN POTTERY
(Frequency of Rim Types by Levels)

TAGA
Rim Types
LEVEL Type A Type B Type C Type D
Stirtacey see ee at 2 Ba S65
Beveltngs.2 as 6 5 1 1
Wevelisi- escesacee 18 14 <3 ae
Wevelesss ee 30 2,
Levelv4ia: wii 32 wh
evel oxo 10 1
evelkGs5.n0 oe 8
evel ey ane 2
WeveliGra7.5 ales 2
WevelbOne es |
MOtalnkt 110 24 1 1
LAULAU ROCK SHELTER
Rim Types
LEVEL Type A Type B Type C Type D
Surtaceaa:.. ce iar a Me Ne
Revelole 5 so 22 8 He 2
eeveli2 any aac 16 5 6 1
Wevelt3 ern aa 24 2 3
eveliaens ers cde 13 es 4
Eevelistne seo 8
Iheevel Grats oeivec2 6
evel ie esas
Levelseei ee
MR Otal epee: 89 15 13 3

Rock Shelter are early, Laulau House A and Blue I are late. In such case,
a high frequency of Marianas Plain Type A rims is characteristic of early
sites, and conversely a high frequency of Type B rims of late sites.

This leaves four sites between the extremes: Objan, Chalan Kija, Oleai,
and Taga. Objan is a latte site and one would expect it to show a higher
frequency of Type B. Nor at Objan was there a significant difference in the
relative frequency of Type A and Type B sherds among the various levels.

This same lack of significance of relative frequency in depth was true of
Chalan Kija and Oleai. At Taga, however, a difference did occur, with the
Type B rims concentrated in the upper levels. As these upper levels are
associated with the Taga latte, the latte period at Taga shows a higher pro-
portion of Type B rims than figures 60 and 61 indicate and brings it closer
to Objan. The data in depth for Taga and for Laulau Rock Shelter—the

POTTERY 127

only two sites with significant depth differences in rim types—are given in
the following tables.

Therefore, within the Marianas Plain category, the difference in rela-
tive frequency of Type A and Type B rims provides supporting evidence in
setting the sites excavated into a relative chronology, though primarily in
regard to sites at the extreme ends of the chronology.

SHERD FREQUENCIES BY SITE

OBJAN
Level
x A se |
Surface 7 2 3 4 5 6
Marianas: Plain’ 2 1255+: 246 992 1,158 801 304 214 148
Marianas Cord-Marked.... ... 5 2 ats nae wists tee
Marianas; Trailed- 3)... 2... Bon Oy cree a ent eet 1
otal. asec era 246 997 1,160 802 304 214 148
Level Seaward Trenches,
- A ~, XX’ pro- shafts
7 8 9 70 file SE ey 12s Total
Marianas Plain. 2 53.3232 303 82 47 63 15 643 935 5,648
Marianas Cord-Marked.... .. ae re oe ee oa 7
Marianas Trailed...:.....: ta = oe ed wae ate. 1
sRotales see yee oe 82 47 63 15 643 935 5,656
LAULAU ROCK SHELTER
Level
eX
LAT 5
Surface 7 2 3 4
Marianas Plaine. ..2o0750 12 324 266 262 129
MarianassRedui tc 6scer = 34 ape: 1 s 8
MarianasReda... circa = ee. ele oor 4
(variant)
Lime-filled Impressed. .....
(trade ware)
Otal 25 ae cae eee ale 12 324 267 267 141
Level
oF
ts = ‘N
uA 6 vA 8 Total
Marianas: Plain so. ses 51 56 2 1 1,103
Marianas: Reds oe 36 31 505 a 81
Marianas Redisssce ee ee. 5 4 oe 0B
(variant)
Lime-filled Impressed. ..... gh 4 i oe 11

(trade ware)

ae
°

2
No)
a
Rey
wm
i)
_—
—
ho
So
a

128

Marianas Plam: 2.3.3.

1 Base of lattes.

Marianas Plain. <: Sa20<.
Marianas (Redii.eaec yan
Marianas: Redtis-7one eee

(variant)

Marianas blames ccna

Marianas Plain’ <2 esac ee
MiarianasiRed’: 2. coexcec
Marianas Rede... 4c

(variant)
Lime-filled Impressed
(trade ware)

Marianas Plain

Marianas Red...
(variant)

Lime-filled Impressed
(trade ware)

Marianas Red) 22.2.5...

MARIANAS PREHISTORY

LAULAU HOUSE A
Level
Ga ~
Surface 7 Hs 3 41 Total
324 1,474 94 84 1,976
OLEAI
Level
| ee — aa
Surface 7 2 3 4 Total
26 105 130 30 5 296
Bas 5 3) a hs 6
1 Ds 1
26 109 133 30 5 303
CHALAN KIJA
Level
r iy tae
Surface 7 2 4) 4 Total
148 178 78 8 1 413
CHALAN PIAO
Level
ae “ aN.
Surface i v4 3 4 5 6
27 407 99 28 12 1 4
1 34 31 48 42 26 24
ee 6 4 16 11 16
ae 1
28 442 136 80 70 38 44
Level
( A cee Y
7. 8 9 70 77 12 Total
eae 6 1 1 Pe en 586
Te: 41 13 6 a 1 397
30 55 itl 2 1 te 152
1
107 102 25 9 1 1 1,136
TAGA
Level
fam ~ ‘
Surface ik 2 3 4 5: 6
27 258 348 229 292 78 24
4
2 258 348 233 292 78 24

POTTERY 129

Level
’ ars A” a
7 8 9 70 77 Total
Marianas: Plain...253% chasse: 5 6 2 4 2 1,275
Marianas Cord-Marked.... rs “i oe Er 4
sBotalt sos siete 5 6 4 2 1,279
BLUE I
Level
F woe A— —\
Surface 7 74 3 4
Marianas Plaine o...0. i123, 1,077 2,402 1,742 777 305
Marianas Fine-Line Incised. Dim gas eps oe 1 ae ae
Marianas Cord-Marked... . 4 5 8 6 2
Marianas Trailed.......... I oe ie ete i ae he
MarranasmeeGs esas ects a a. nee are iene Aen Ranta
SPotal osha sates 1,085 2,405 sr Asy. 783 307
Level
| es A ™~
5 6 7 8 9 Total
Marianas: Plainic hice 290 109 108 35 42 6,887
Marianas Fine-Line Incised. Rae aR oe te te 3
Marianas Cord-Marked.... 1 Sean se tt oe 24
Marianas) Tratled 2). ca eacs es Se ees i eee 3
Marianas Reds .caak ec Pee Ree 1 AS ae 1
otal. eee waver cores 291 109 109 35 42 6,918
ROTA SURVEY TINIAN SURVEY
Surface Surface
MarianastPlainl snes osteo 100 Marianas: Plain: cia ante 862
Marianas Fine-Line Incised...... 4 Marianas Cord-Marked......... 1
Marianas Cord-Marked......... 18 Marianas “Crasledt 0. cath 2
Potalent meee ook ee ees 122 Ota cre hetae soe eee ee 865
SAIPAN SURVEY
Surface
Marianas-Plain.-400 seo 68

POTTERY AND BAKED CLAY ARTIFACTS

A few pottery and baked clay artifacts, other than sherds, were obtained
in survey and excavations.

Sinker.—Small, ovate sinker with pointed ends and with groove around
mid-section. Length 37 mm. Found at Laulau Rock Shelter, Level 5 (see
fig. 62, upper left).

Pendant.—Large circular pendant of Marianas Plain pottery, probably a
worked sherd, with perforation near edge of disc. Surface find at Chalan
Galeite, Saipan. Diameter 67 mm., thickness 21 mm. (see fig. 62, bottom).

Pendants or Sinkers—Two circular worked sherds of Marianas Plain pot-
tery, with perforation in center. (1) Whole specimen found at Objan,

130 MARIANAS PREHISTORY

Level 6; diameter 35 mm. (see fig. 62, upper right). (2) Fragment found at

Blue I, Level 2; diameter 31 mm.
Pottery Smoother(?).—Large, oval object with rounded ends and one flat

longitudinal surface. Surface find at Muchon Point, Rota. Length, 87 mm.

Fic. 62. Pottery and baked clay artifacts. Upper left, sinker; upper right, worked
sherd pendant; lower, pendant.

VI. Stone Artifacts

ADZES

Ten whole adzes and eleven fragments were recovered from Saipan,
Tinian, and Rota. This collection is so small that few significant conclu-
sions can be drawn from its examination. However, the very paucity of
stone adzes in comparison with shell adzes, particularly on Jatte sites, is
interesting. Although the prehistoric Chamorros were fully conversant
with pecking, grinding, and polishing techniques as applied to stone, for
material for their adzes they relied primarily on Tridacna shell. The latter
was more easily worked and the material more readily at hand, for most of
the igneous rock of which Saipan’s and Tinian’s stone tools were made was
probably imported from Anatahan and other northern islands. In connec-
tion with the paucity of stone adzes, it is true that /atte sites are generally
shallow, that there has been much artifact-collecting in the Marianas, that
probably most sites have been worked over by collectors, and that many
stone adzes have been removed. Nevertheless, I believe that on Jatte sites
shell adzes have always outnumbered stone ones.

The stone adzes have been classified according to three types. Types 1
and 2 correspond to Thompson’s Types 1 and 2 (Thompson, 1932, pp.
33-36), except that some of the rectangular adzes included in her Type 2
I would classify separately and would place with my Type 3. Thompson
also describes a Type 3, but the brevity of her description makes the identi-
fication of her Type 3 adzes very difficult when her classification is applied
to a new collection.

The three adze types established on the basis of my small collection are
described in the following pages. The conclusions given below have been
drawn from examination of these types:

(1) One cylindrical adze of Type 1 was found in Level 9 (4.0—4.5 feet
below ground surface) at Chalan Piao, Saipan, in association with Mari-
anas Red sherds. If the carbon 14 date of 1527 B.c. + 200 for Chalan Piao
is correct, this is the oldest adze in the collection. Type 1 adzes were also
found at Objan, a Jatte site, the upper level of which is probably historic.
This evidence, therefore, indicates that Type 1 adzes persisted for a very
long period of time.

131

132 MARIANAS PREHISTORY

Fic. 63. Stone adzes, Type 1. Provenience of specimens as follows: upper, Marpo
Valley (Tinian); center, Objan (Saipan); lower, Chalan Piao (Saipan). Length of top
specimen, 133 mm.

(2) At latte sites, Type 1 and Type 2 adzes coexist, both as surface and as
subsurface finds.

(3) Type 2 adzes are clearly similar to shell adzes of the same shape.
The same form is therefore expressed in both shell and stone.

(4) The two Type 3 adzes represent surface finds and no conclusions are
possible regarding their age.

STONE ARTIFACTS 133

Description of the adze types follows. The nomenclature used is taken
from Buck, Emory, Skinner, and Stokes (1930). One adze praRmen was so
small it could not be classified.

STONE ApzEs, TYPE 1

Shape.—Cylindrical in front-to-back silhouette, tapering toward both
poll and cutting edge. Poll round or flattened. One longitudinal surface
may be slightly flattened, presumably to facilitate hafting. Chin rounded.
Adzes exhibiting the best workmanship belong to Type 1.

Cross Section.—Circular or slightly elliptical at mid-section. The best-
made adzes are circular. Figure 63 (center) illustrates a broken fragment.
The cross section of this fragment is elliptical, but it was taken from below
the mid-section.

Surface.—The surface has been pecked. The blade has then been care-
fully ground, and the butt, particularly the poll, may exhibit light grinding
over the pecked surface. In some specimens, the area of the surface adjacent
to the cutting edge is polished.

Bevel and Cutting Edge-—Cutting edge may be straight or curved. In
front-to-back silhouette, the cutting edge varies from a slight to a pro-
nounced downward curve. The bevels are arched and confluent with the
back. In figure 63 (upper specimen), the margins of the bevel have been
ground to the point where they are virtually indistinguishable. The margin
of the bevel is indicated by a dotted line.

Size.—Length of whole adzes varies from 9.1 to 13.3 cm.

Remarks.—Two specimens have cutting edges battered through use, and
reground to form blunt, maul-like edges.

Provenience.—(a) Four whole adzes. Saipan: Chalan Piao (Level 9);
Objan (Level 2, Level 5). Tinian: Upper Marpo Valley (surface). (b) Six
fragmentary adzes. Saipan: Objan (surface, Level 4). Tinian: Blue I (sur-
face); Blue VIII (surface). Rota: Muchon Point (surface).

STONE ApzEs, TyPE 2

Shape.—Approximately rectangular to triangular in front-to-back sil-
houette. Surfaces rounded and angular margins absent. Poll rounded,
occasionally slightly flattened. Front, plane or slightly concave; back,
convex. There is no pronounced chin.

Cross Section.—Flattened ellipse at mid-section.

Surface.—Allover pecked surface. The amount of subsequent grinding
varies. Most specimens exhibit some grinding over the whole adze surface,
in addition to the bevel and edge.

Bevel and Cutting Edge.—In silhouette, cutting edge is convex. Bevel is
sometimes slightly concave, forming a gouge-like tool. This latter feature is

Fic. 64. Stone adzes, Type 2. Provenience of specimens as follows: upper, Muchon
Point (Rota); center and lower, Objan (Saipan). Length of top specimen, 116 mm.

134

Fic. 65. Stone adzes, Type 3. Provenience of specimens as follows: Saipan, south
section of island, no site. Length of top specimen, 178 mm.

135

136 MARIANAS PREHISTORY

pronounced in one specimen (fig. 64, upper) which also has a secondary
bevel, indicated in the drawing by stippling. The variation in the amount
of beveled area is illustrated (fig. 64).

Size-—Length of whole adzes varies from 7.6 to 11.6 cm.

Remarks.—Cutting edges of two specimens have been battered through
use and then slightly reground. Polls also exhibit battering. These adze
blades may have been re-used as wedges.

Provenience.—(a) Five whole adzes: Saipan: Objan (Level 3); Laulau
House Site (Level 1). Tinian: Blue II (surface). Rota: Muchon Point (sur-
face). (b) Four fragmentary adzes: Saipan: Objan (Level 3, 4); surface
find (no site). Tinian: Marpo III (surface).

STONE ApzEs, TYPE 3

Shape-—Roughly rectangular, in front-to-back silhouette. Flattened
polls. Front and back surfaces plane. Sides rounded. The only two speci-
mens of this type in the collection exhibit crude workmanship. One adze is
triangular in profile (fig. 65, upper specimen).

Cross Sectton.—Rectangular, with curved sides.

Surface.—The two adzes of this type exhibit both chipped and pecked
surfaces. Except at bevel and cutting edge, grinding has been used only
very slightly over the adze surface.

Bevel and Cutting Edge.—Beveled surface narrow. Cutting edge straight.
On one adze (fig. 65, upper specimen), only the bevel is ground. On the
other specimen, both surfaces of the cutting edge are ground.

Size.—Length varies from 12.6 to 17.7 cm. These are the largest adzes in
the collection.

Provenience.—Saipan: South section of island, no site (isolated surface

finds).
CHISELS

One whole specimen and one fragment have been classed as chisels, pri-
marily on the basis of their small size (fig. 66). They are cylindrical in
shape, resemble stone adzes of Type 1, and conform to Thompson’s de-
scription (1932, p. 37). Description of the single whole specimen follows.

Shape.—Cylindrical in silhouette, with sides tapering toward poll and
cutting edge. Poll flattened. No pronounced chin.

Cross Section.—Circular at mid-section.

Surface.—Ground on all surfaces.

Bevel and Cutting Edge.—Bevel surface elliptical in outline and confluent
with the back. In silhouette, cutting edge is convex.

Size.—Length 7.6 cm.

Provenience.—Muchon Point, Rota (surface find).

STONE ARTIFACTS i Be

Fic. 66. Stone chisels. Provenience of specimens as follows: upper, Muchon Point
(Rota); lower, Blue I (Tinian). Length of top specimen, 76 mm.

GOUGES AND AXES

Buck, Emory, Skinner, and Stokes (1930, p. 179) define a gouge as “‘a
special form of chisel in which the edge is curved to such a degree that the
bevel is hollow or grooved.” No special forms of chisels conforming to this
description were found. However, one adze of Type 2 (fig. 64, upper) does
have a gouge-like bevel and edge.

No specimens that could be definitely classed as axes were found, either
as surface finds or as excavated specimens. One artifact from the surface at
Objan may represent an ax fragment, but the classification is uncertain.

SLINGSTONES

Thirty-five slingstones were found at sites excavated or surveyed. This is
a small series compared to the 4,702 in the collection made by Hornbostel
some thirty years ago. Slingstones are to the Marianas as projectile points
are to the United States, and the surface of Guam, Tinian, and Saipan has
been well scoured by collectors.

From documentary accounts, it is clear that the sling was a principal
weapon of the Chamorros. The usefulness of slingstones as time-indicators,
however, remains to be demonstrated. The potential variation in form is
not great enough to provide a clear-cut medium of expression for stylistic
change. Slingstones vary in quality of workmanship (which may be purely
random), in shape (primarily whether the ends are pointed or rounded),
and in size. In the following classification, three slingstones with rounded

138 MARIANAS PREHISTORY

ends have been set aside as a separate type; the remainder are grouped to-
gether. The three with rounded ends are all from Saipan, but this is not
meant to imply that the type does not exist on the other islands.

Thompson (1932) classified the Hornbostel series into four types. She
noted the presence on Guam of slingstones of baked clay and observed that
local variation in slingstone materials correlates with the geology of Guam.
No baked clay slingstones from Saipan or Rota are represented in the
Hornbostel collection (Thompson, 1932, p. 100), and none were found in
my own. Thompson’s comment on local variation of slingstones on Guam
is the only useful generalization to which discussion of slingstone types has
contributed.

It should be noted that one slingstone from Saipan is 109 mm. long—
nearly twice the usual size. The Hornbostel collection contains a number
of oversize stones, the largest 260 mm. long. A ceremonial use for these
oversize stones has been suggested (Thompson, 1932, p. 51).

SLINGSTONES WITH PorInTED Enps (Fig. 67)
Dimensions.—Length 44-109 mm.; average 58.6 mm. Diameter 15-51
mm.; average 29.9 mm.

Material.—Coral limestone.

Provenience
SAIPAN TINIAN

Objan: Blue I:

DUriACe matic nce he Surfaces aos sss tec 1
XOX prOniles. «cater 1 bevelezeeens tay 1
evel es oho 4 evele3e a, ekacae oe 1
héveluSiecninace sexe al evelobiese sy asc ean ail
Weve liana micas iyi 2D,

Laulau House A: Tinian Survey:
Surtacetacas eer lives Uilee se ene it
evel el eur et one oT IBIUCSV/ See ee 1
eve lia as sack tex 1 Puntan Laminbot 1

Laulau Rock Shelter: Motalesunsesesatieke Ui
Weevelcdsscesey. mation 1

Oleai: ROTA
SUIACE Hara nee meee ee Survey:
eVele2e are ea on 1 Muchon Point...... 4

Chalan Kija: OUD) techies whe: 32
Surlace ns eee 1

Saipan Survey.... 1

‘uu ¢¢ ‘usuTdeds do} Jo 4 Sua'T “UI 6()] SUoUT
‘spu9 popuNOL YIM souOysSuTTG “gg “OI -lpads ysomo] Jo yISuaT ‘spus pojurod yim souoysSurg “9 ‘Ory

19)

140 MARIANAS PREHISTORY

SLINGSTONES WITH RouNnDED Enps (Fig. 68)

Dimensions.—Length 55-64 mm.; average 58.7 mm. Diameter 2.9-3.9
mm.; average 34 mm.
Material.—Coral limestone.

Provenience
SAIPAN
Objan: Laulau House A:
Surfaces: - in csonenes 1 Surtacenw:euoncnecvaes el
Weveli4 uses a accel a
MOLAe maeaieccare <6 5
PES LLL

The Chamorros at the time of the first Spanish contact were rice
growers, and heavy stone mortars, presumably for grinding rice, are com-
monly found in association with J/atte sites. ‘The stone pestles found at
archaeological sites are of two general types: (1) relatively long cylindrical
forms (fig. 69); and short, stubby pestles (fig. 70). The former could have
been used with the larger, heavy mortars. The stubby forms were probably
used for grinding medicinal plants in smaller stone mortars, a use to which
they are still put today on Saipan. All pestles found were associated with
latte sites.

RELATIVELY Lonc CyLinpRIcAL PEsTLEs (Fig. 69)

Dimensions.—Length 129-184 mm.; average 155.5 mm.
Material.—Igneous rock.

Provenience
SAIPAN
Objan: Chalan Piao:
DUTTACE ar ote eons 1 Surtaces soca. ol
Heveligyc wert 1
Saipan Survey: Afetna:
Fanunchulujan..... ik Surlace yo cs 1
Oleai: Location unknown:
Suriace as see eel Subface +) ace: 1
TINIAN
Blue I: Tinian Survey:
SULLACER Mere aeas eee ge Blues ere se 1

Leprosarium I...... 1
Unai DangkuloIII.. 1

STONE ARTIFACTS 141

Fic. 69. Pestles. Length of upper left
specimen, 141 mm.

Fic. 70. Pestles. Length of upper right
specimen, 111 mm.

SHORT, StuBBy Pest.es (Fig. 70)

Dimensions.—Length 56-111 mm.; average 80.3 mm.
Material.—Igneous rock.

Provenience
SAIPAN TINIAN
Objan: Blue I:
SUMACE ence eel he 1 Suntacea ee ae 1
PeveliGucc steer 1
Miarpos tli tricc tr: 1
Laulau House A:
Surlace cies reel ROTA
Sljangigani.wee cee,

UNCLASSIFIED FRAGMENTS

Dimensions.—Length 53-110 mm.; average 77.6 mm.
Material.—Igneous rock.

142 MARIANAS PREHISTORY

Provenience
SAIPAN ROTA
Objan: Muchon.Point....2::. <1
eveletie sy -aik.e a pane 7 od =
Wevelia Aes ee
TINIAN
Blue I: Tinian Survey:
Surfaces. fp nesne 4 Bluee Vil sass 1
Motalle Wn hase: 9
MORTARS

Large, igneous boulders used as mortars were frequently found in associ-
ation with Jatte sites. Others occurred as isolated finds, but it is obvious that
many have been displaced and moved to new locations, often as recently as
World War II. The non-latte sites excavated—Laulau Rock Shelter, Oleai,
and Chalan Piao—possessed no mortars, but in view of the extent of dis-
placement this has no significance. Mortars were, however, found appar-
ently undisturbed at atte sites and I believe they are particularly associated
with these sites. All mortars were discovered as surface finds. None were
found buried beneath the surface.

The most logical sources for these mortars are the rocky beaches of the
volcanic islands north of Saipan, as neither Saipan, Tinian, nor Rota pro-
vides basaltic boulders of this type. ‘They apparently were moved to the
southern islands by the ancient Chamorros.

The boulders, though sometimes circular in form, were apparently un-
worked; the rounding of edges probably was caused by water action. One
possible exception—a mortar found at As Teo—was in the form of a nearly
perfect disc.

Although Hornbostel reported mortars with as many as seven mortar
holes, only one multi-holed mortar was found during my own work. It was
discovered at Unai Bapot on Saipan and possessed three mortar holes. All
others were single mortars, whose grinding depressions ranged in top diam-
eter from 0.5 to 0.9 feet (average 0.72 feet), with depths ranging from 0.4
to 0.8 feet (average 0.52 feet) (figs. 71 and 72).

SMALL STONE MorTARS AND STONE CUPS

The Hornbostel collection described by Thompson (1932) contains a
number of small stone mortars, as well as stone cups. None of these were
found during my own survey and excavations.

STONE ARTIFACTS 143

MAULS

One maul was found on the surface at Objan. It is of basalt, with pecked
surfaces, and with both horizontal and vertical grooves for lashing and
hafting (fig. 73). Length 124 mm.

Fic. 72. Stone mortar in situ, Laulau
Fic. 71. Stone mortar found on Tinian House A, Saipan.
near the House of Taga. This mortar has
been displaced from its original position.

HAMMERSTONES

Two spherical hammerstones (fig. 73) and a fragment of a third were
surface finds on Saipan and Tinian. One other hammerstone, with flat-
tened sides and pecked depressions on both sides, presumably to facilitate
grasping the stone, was found on Tinian (fig. 73). The surface of Objan
also yielded three fragments which may be hammerstones but which could
not be classified with certainty.

SPHERICAL HAMMERSTONES

Dimensions.—Diameters of whole hammerstones: 62 XK 82 mm. and
121 XK 142 mm.

Provenience
SAIPAN TINIAN
Objan: Puntan Laminbot:
Surlaces.c 738g: st: 1 DUIIACES . yeeee pat na ead
Fafunchulujan: Potaby wie av ats i

144 MARIANAS PREHISTORY

Fic. 73. Upper left: pestle. Upper right: maul. Center and lowest specimens:
hammerstones. Center right specimen has pecked depression on both sides. Length
of upper right specimen, 124 mm.

NotTcHED HAMMERSTONE

Dimensions.—Length 114 mm.

Provenience.—Tinian, Blue I, surface, 1.

SINKERS

Fishing sinkers take a variety of forms. Some are merely reworked sling-
stones; others are more specialized. Materials used include both limestone
and igneous rock.

STONE ARTIFACTS 145

Fic. 74. Hemispherical sinkers, viewed
from above. Diameter of upper specimen,
102 mm.

Fic. 75. Pyramidal sinkers with round-
ed surfaces. Groove for lashing shown on
lower specimen. Length of upper speci-
men, 89 mm.

HEMISPHERICAL SINKERS

An account of the use of this type of sinker for fishing was collected by
Hornbostel from Chamorros on Rota. The account is included in Thomp-
son (1932, pp. 47-48). Two specimens were found (fig. 74).

Dimensions.—Diameter of base 92 mm. and 102 mm. Height 69 mm. and
82 mm.

Material.—Coral limestone.

146 MARIANAS PREHISTORY

Provenience.—Both specimens are surface finds from Muchon Point,
Rota, in association with Jatte structures.

PyRAMIDAL SINKERS WITH ROUNDED SURFACES

This sinker type is almost conical in form, except that two opposite sides
are broader than the other two sides, forming a rectangular base. However,
all edges have been pecked away to form rounded surfaces. Longitudinal
grooves, approximately 5 mm. wide and from 1-3 mm. deep, have been
pecked along the midline of the opposite sides. Out of four specimens, the
grooves are on the broad sides in two specimens and on the narrow sides in
the two other specimens (fig. 75).

Dimensions.—Length 81, 89, 105, 114 mm. Dimensions of base: width
64-95 mm.; thickness 54-73 mm.

Provenience
SAIPAN TINIAN
Objan: Blue I:
Siirlacés trae i ee ek Stntace sac atereeet 2
Thevell ot ieee a 1

MIscELLANEOUS SINKERS

These consist of five small sinkers (fig. 76). Three are apparently re-
worked slingstones, with grooves cut either longitudinally or transversely
into the stone. One of these three has one point of the slingstone worked off.
A fourth sinker (fig. 76, center) may also be a slingstone. At the top of the
sinker, two small holes which meet within the matrix have been bored to
allow suspension of the sinker from a fishing line. The fifth sinker is spheri-
cal, with a narrow groove cut around one circumference (fig. 76, lower
left).

Dimensions.—Longest dimension varies from 41 to 55 mm.

Material.—Coral limestone, igneous rock.

Provenience
SAIPAN
Objan: Afetna:
Surlace. sis ese 1 Sumlaces crane 1
Oleai: Location unknown:
Sublaces = sae ae ee 1 Slirtacewe acetate & 1

Chalan Piao:
SUurlace ets tra eee 1

STONE ARTIFACTS 147

Fic. 76 (left). Sinkers. Upper left, upper
right, and probably lower right, re-worked
slingstones. Center specimen probably a re-
used slingstone of the rounded point type.
Length of upper left specimen, 54 mm.

Fic. 77. Pendants or sinkers. Top speci-
men shell, lower two” specimens stone.
Length of top specimen, 65 mm.

PENDANTS

Three pendants of a typical Marianas form, numerous in the Horn-
bostel collection, are illustrated (fig. 77). These may be sinkers; there is no
ethnological or documentary evidence to determine the point. Two of the
specimens are fragments. The one whole specimen is actually of shell, prob-
ably Tridacna, but the form is identical with the two fragmentary stone
specimens. A hole for suspension has been bored at each end of the speci-
men.

Dimensions.—Length of whole specimen 65 mm.

Material.—Igneous rock and shell.

Provenience
SAIPAN TINIAN
Objan: Blue I:
eve An ise cc 1 SUTlAce cere eee 1
(shell specimen) Reveli2nt. stants ie 1
WV OtAL. eestor 3
SCRAPERS

CIRCULAR SCRAPERS

Perhaps the most distinctive scraper form we found is roughly circular,
in the shape of a disc, usually with some retouching around the edge. Left

148 MARIANAS PREHISTORY

Fic. 79. Upper three specimens: prob-
Fics 7 Socapers, SMapsepecimens care ably knives. Lower specimen: large knife or
‘ : possibly a scraper. Length of lower speci-
scrapers. Bottom three specimens: circular
: en, 140 mm.
scrapers. Length of top left specimen,
34 mm.

middle and lowest specimens in figure 78 illustrate the type, which future
excavations may show to be common in the Marianas.
Dimensions.—Diameter 67-105 mm.
Material.—Basalt.

Provenience
SAIPAN TINIAN
Objan: Blue I:
evel Qe ke siete: 1 Tee vel yah ee seghes a cacs al
Leprosarium I:
Suriaces.cssvosrmdes 1
MMotali 3 ees rotaes i)

In addition, a much smaller circular scraper (diameter 35 mm.), made

from quartz, was found on the surface at Objan. Edges have been re-
touched (right middle specimen, fig. 78).

STONE ARTIFACTS 149

Fic. 80. Unworked cores and flakes. Lower two specimens of volcanic tuff.

SIDE SCRAPERS
Two flakes, probably side scrapers, are shown as the upper specimens in
figure 78.
The upper right specimen shown in the figure exhibits some secondary
flaking along the working edge. In addition to these specimens, a number

of unworked flakes were found that may have been used as scrapers (fig.
80).

Dimensions.—Length 42-85 mm.
Material.—Basalt.

150 MARIANAS PREHISTORY

Provenience
SAIPAN TINIAN
Objan: Blue IV:
Surfacexmi siete eal Surfaces oes 1
MM prone eee 1 -——

PhOtale oe waste hoe go

END SCRAPER

One end scraper was found at Blue I, Level 3. It is 41 mm. long, of
basalt, with a slight amount of secondary flaking along the working edge.

KNIVES

A few flakes found are probably knives (fig. 79). The lowest specimen is
much larger and may have served as a scraper.

Dimensions.—Length 60-140 mm. Width 38-78 mm.

Material.—Igneous rock.

Provenience.—Objan (Saipan): Surface, 2; Level 1, 1; Level 3, 1. Total 4.

UNWORKED CORES, FLAKES, AND STONE FRAGMENTS

Every excavated site yielded a small collection of unworked basaltic
flakes and small cores (fig. 80). As noted previously, unworked flakes may
have been used as scrapers. A feature of almost every /atte site was the find-
ing of a few fragments of volcanic tuff. Their use is unknown, as they are
unsuitable for cutting instruments. Waterworn basalt beach pebbles, pos-
sibly used as pot smoothers or as raw material for flakes, were also occa-
sionally found at latte sites.

VII. Shell, Bone, and Metal Artifacts

SHELL ARTIFACTS
ADZES

As noted in the section on stone adzes, many more shell than stone adzes
were found during the course of excavations and survey. The implication is
that the prehistoric inhabitants of the Marianas relied primarily on shell
rather than stone for cutting tools. However, the shell adzes are small and
it is probable that stone adzes were used for heavy work, and shell tools for
lighter work and for finishing. Also, the shell adzes were predominantly
made from Tridacna, plentiful in the coastal waters of Saipan, Tinian, and
Rota. On the other hand, the hard, igneous rocks needed for stone adzes
had to be imported, presumably from the Mariana islands to the north.

In describing the shell artifacts from the Hornbostel collection, Thomp-
son distinguishes between Tridacna shell adzes and Tridacna scrapers, de-
spite the fact that the forms as she describes them are virtually the same.
She notes that “‘scrapers of Tridacna differ from the prevalent shell adze
type in that they are thinner and tend toward greater longitudinal and
edge curvature.” (1932, p. 54.) Additional definitive information for clas-
sifying these artifacts as scrapers is not given. In this report they are classed
as adzes.

Thompson classifies the shell adzes of the Hornbostel collection into
three types. Her ““T'ype 1” includes all the Tridacna shell adzes described in
this report. Her ‘‘Type 2” shell adzes are identical in form with the cylin-
drical stone adzes described as ‘‘Type 1”’ stone adzes in both Thompson’s
and my reports. No shell adzes of this type were found in my own field
work. Thompson’s ““Type 3” shell adzes are large Caroline Island forms,
probably imported in historic times. This type is not represented in my own
collection.

Shell adzes found during the Museum’s work on Saipan, Tinian, and
Rota are of three types: triangular; quadrangular (those with opposite
edges roughly parallel); and trapezoidal, according to front to back outline
(fig. 81). With one possible exception (fig. 81, top center) all are made of
Tridacna. The polls of triangular adzes are often rounded, so that the tri-
angular shape is much modified. Edges of all adze types are often battered,

151

152 MARIANAS PREHISTORY

Provenience of Triangular Shell Adzes

Level
Sur- XX! = A
face profile 7 2 3 4 a Total
Saipan:
O©bjanten sea aarcweacon 2, 1 2 1 2 1 9
Laulau Rock Shelter... ee 1 ae ne Me oe fl
C@halan.Riadve xn nie ae ot, iM 1 .y ie , 1
Oleait eos ee oe 1 A eae ra cx mts a 1
AS EOL aan enn ees 1 1
Location unknown... 1 1
Tinian:
Bluedle an Sasa 3 4 iE
PAG AT ee tate opt cas tes 1 1 2.
Ble Ese 1 1
BIG 2 Vata rast ere we ears il 1
BlWe sXe es ete 1 1
eprosarium ace seo 1
Iheprosarium Uljss...- 1 1
Rota:
Muchon Point...... 1 ae we ro ioe La nie 1
Motale a b4e cs aoe 2 2 pe, 8 2 2 m1 29
Provenience of Quadrangular Shell Adzes
Level
BOL oat A =
Surface profile 7 2 6 8 Total
Saipan:
Objanti see wees. 1 1 1 2 1 Al a
GhalansPiaoe jee 1 ko 1 f : Le 2,
Oleaic ics eee 1 1
ASR COG teen crip aes 1 1|
Tinian:
Blues sas ees oie 2 1 3
Leprosarium I... . 1 1
Totaly noe ih 1 2 3 1 1 15
Provenience of Trapezoidal Shell Adzes
Level
_———
Surface 2 4 Total
Saipan:
@Objanie sone sears ee 1 1
Tinian:
Bless Bac rete ee oe 1 _ 1
Sap aan tee siete teers vy teg saseees Ss tes 1 1
ILeprosaniumiuls.. 3 .c1s2ee0 1 ah aes 1
“otal pitti Gee tie 1 1 2 4

but as Thompson notes, grinding of the bevel is primarily on the back side.
On the basis of present evidence no stratigraphic difference can be as-
cribed to the three types set forth here, and the typology proposed here is
merely a matter of convenience in description.

Fic. 81. Shell adzes and gouges. Length of center specimen in top row, 107 mm.

153

154 MARIANAS PREHISTORY

Shell adzes in the Museum collection range in length from 62 to 120 mm.
(average 82.7 mm.).

UNCLASSIFIED ADZE FRAGMENTS

Eleven fragments of Tridacna adze blades could not be classified. All
these fragments include the cutting edge, plus from 4 to 7 mm. of the blade.
Apparently, while in use shell adzes tended to break transversely above the

edge.
Provenience of Shell Adze Fragments
Level
XX = A =e
Surface profile i 2 7) 6 Total
Saipan:
O©)Djan ieee woe ee: 1 ae 1 1 ey 5
Laulau Rock Shelter .. ate Ve 1 a 1 2
Chalan-Piaow.. 4. 1 Re res ae a 1
Tinian:
Blue ses ye eta be 1 11
Ma Galeria mo calsressiens 1 1 1
Rota:
Muchon Point... .. il 1
otaleiacwe eee 3 1 1 3 1 2, 1A

ADZE BLANKS

Two partially worked Tridacna adze blanks were found, one as a surface
find at Objan and one in Level 3 at Blue I. Length of blanks 81 and 88
mm.

GOuGES

Only one complete shell gouge was recovered (lowest center specimen,
fig. 81). This is made from Tridacna and was a surface find at As Teo
(Saipan). In addition, two other artifacts, probably gouges, of the same
long, thin form were obtained. They are not complete specimens, as the
cutting edges have been broken off, so it is not possible to classify them
with certainty. Finally, one fragmentary gouge of Terebra was also found
(lowest right specimen, fig. 81).

Material Length Provenience
Gouge...... Tridacna 90 mm. As Teo (Saipan), surface
Gouge(?).... Tridacna 77 mm. Objan (Saipan), Level 1
Gouge(?).... Tridacna 80 mm. Sijangigani (Rota), surface

Gouges ss herebra 41 mm. Objan (Saipan), XX’ profile

SHELL, BONE, AND METAL ARTIFACTS 15:

ul
vi

PEDALION KNIVES OR SCRAPERS

A considerable number of thin flakes of Pedalion found in the excavations
may have been used as knives, scrapers, or even spoons. Their use is uncer-
tain. Four typical specimens are illustrated (fig. 82). Length of whole
specimens ranges from 60 to 95 mm.; average 68.4 mm.

Fic. 82. Pedalion shell knives or scrapers(?). Length of upper left specimen, 95 mm.

156

Fic. 83 (above). Turbo shell scrapers.

Fic. 84 (left). Shell container for lime
chewed with betel nut; shell perforated
for suspension.

SHELL, BONE, AND METAL ARTIFACTS 157

Provenience of Whole and Fragmentary Specimens

Level
‘ a7 *~ Faz,
Surface 7 Z 3 4 | 6 7 Total
Saipan:
Objans25 sc 11 1 17 4 1 3 1 38
@Oleatne2s 1 as ae put 5 ac cs fe 1
Tinian:
Blue. ce 2 6 Or 8
WV aBansts a) ty Fee x 1 1
Total. . 1 1S 7 18 4 1 3 1 48

TuRBO SCRAPERS

A series of Turbo shell scrapers, similar to those known ethnologically
from different parts of Micronesia, was all, with one exception, found at
the Objan site on Saipan. Two of these scrapers are shown (fig. 83).

Provenience
Level
if te za:
Surface 7 2 3 6 8 Total
Saipan:
Objans ase oeca ee 3 4 8 1 1 19
Fafunchulujan... 1 Sig ave ne oo 1
UGtal i350 s5:22 os 3 3 4 8 1 1 20

LimE CONTAINER

From ethnological evidence, and the fact that the teeth of ancient buri-
als are often stained a dark brown, it is certain that the modern Chamorro
custom of betel-nut chewing has a very respectable antiquity. Betel nut is
commonly chewed with lime, which releases the narcotic in the nut. Figure
84 illustrates an ancient shell lime container, found on the surface at Objan
Site on Saipan. A deposit of lime remains in the shell, and the deposit itself
contains a well-delineated finger impression. The shell container is perfo-
rated, probably to accommodate a cord hung around the neck of the user.

Fish Hooxs AND GORGES

With the exception of one fragmentary specimen reported by Thompson
(1932), all prehistoric Marianas fish hooks of which I am aware are of a
single small type (upper row of hooks, fig. 85). The center specimen is the
only complete hook found. It is 28 mm. long and has two small notches on
the upper end of the shank for attachment to the line.

158 MARIANAS PREHISTORY

Provenience
Level
XX! jaa Sat)
: profile 3 4 Total
Saipan:
Objanj.evnc 4 1 1 3

Fish gorges likewise conform to a single principal form (see Thompson,
1932, p. 46). Two gorges are shown (center row, fig. 85). In addition, three
partially completed gorges were obtained, one of which is shown (lowest
specimen, fig. 85).

Provenience
FINISHED SPECIMENS
Surface Level 2 Total

Saipan
Objaneaa.s 1 fs 1

Tinian:
Blue I...... 1 1
Total. 1 1 2

PARTIALLY FINISHED SPECIMENS
Level 7 Level 2 Level 3 Total

Saipan:
Objan yes eves oe ae ne 1 1
Laulau House A...... 1 ne 23 1

Tinian:
Blue I 1 1
Motaler pee 1 1 1 %)

SHELL BEADS AND PENDANTS

A number of circular beads and pendants, perforated in the center and
no doubt used as ornaments, were found in the excavations. They are
roughly of two sizes. The smaller series ranges in diameter from 6 to 25
mm. (average 14.5 mm.), the larger from 32 to 65 mm. (average 42.6
mm.). Four of the smaller specimens (top row, fig. 86) are probably beads.
Two larger specimens (center row, fig. 86) are probably pendants. The
holes in these specimens are characteristically bored from one side only.

Provenience
SMALL SPECIMENS
Level 7 Level 2 Level 3 Total

Saipan:
Objanw ey acstan te i: ay 2 2
Laulau House A...... 1 ae i 1
Tinian:
Blick ss hee eee 1 4 2. ah

‘WU C9 “J9[99eIq [[IYS Jo 19}9UTeIG
‘jusUIeUIO se pasn AlTqeqoid ‘ysn} vssnuigvg jo JUDUIsey

7YSI1 woyWog “joO[90eIq [JOYS :IJe] WoWog *(¢)s}uepuod yisul']
[[9YS :MOI 19}U97 ‘syooY YsyY []IY

124s >MOdI J2}UIT) *(é)spesq TT?y4s :-Mol doy, “O8 ‘OI

“WUIUL 87 ‘MOI do} ul usu1deds I2}UQ9 jo

‘98108 poyojduroo Ayyensed :wo0g

:MOl do J

*s9s108 YSU

“SI

160 MARIANAS PREHISTORY

Provenience
LARGE SPECIMENS
Surface Level 1 Level 2 Total

Saipan:
Objangtiee cdots 1 a 2 S
Laulau House A...... 1 1 Sie 2

Tinian:
Blue I 1 1
Motal eas yee 2 2 2, 6

SHELL BRACELET

One Conus shell bracelet (diameter 65 mm.) was found on the surface at
the Chalan Piao site on Saipan. Thompson (1932, p. 57) reports that a
similar bracelet in the Hornbostel collection was found at a depth of twelve
inches during Hornbostel’s test digging at the Taga Site. Although brace-
lets of this type are known ethnologically from the Carolines, it is probable
that these two bracelets are indigenous to the Marianas.

PENDANT OR SINKER

One shell pendant or sinker has been described in connection with stone
artifacts of the same form (fig. 77, top specimen).

BONE ARTIFACTS
AwWLSs OR THATCHING NEEDLES(?)

Figure 87 illustrates a number of specimens which were probably used
as awls or thatching needles. The four left specimens shown in the photo-
graph have been worked. The fifth specimen, shown at the right, is the
upper mandible of a tern; whether or not it was used as an artifact is not
known. From left to right the specimens are:

Specimen Length Provenience

Awl(?) 123 mm. . Laulau House A, Level 1 (Saipan)
Awl(?) 118 mm. Laulau House A, Level 1 (Saipan)
Awl 79 mm. Objan, Level 3 (Saipan)

Awl 48 mm. Blue I, Level 3 (Tinian)

Awl(?) 42 mm. Laulau House A, Level 1 (Saipan)

SPEAR PoINTs

At the time of first European contact, Chamorro spears are said to have
been tipped with bone points, made from human long bones. Thompson
(1932, p. 52) reports three such bone spear points from the Hornbostel
collection. Two fragmentary specimens were found at Laulau House A,
Level 1 (fig. 88).

Fic. 87. Bone awls or thatching needles(?). Length of left specimen, 123 mm.

161

Poe

se
ese,

O8ld
ueyeyy)

‘unosnyy ALOISTFY [RINIVN OSKOTY!) ‘so}v1q9}19AUT JOMO'T “‘royeinyD ‘seep Ziq ‘aq Aq paynuepy ;

x ore Bey tie Pits. ewan ve

a a
*

elry (6) Japys = yesnoy urelqo

ureyy YoY nepne'y
nepne'y

ALIS GALVAVOXE WOU TIAHS GAaXYYOMNN

“** * QUUT'T DuDitinDU (DYLND Py) DILEND PW

Sets ss < UTTIUID) o24gsty (DIIGD4p) vYLEND PY

ua ouul'yT DIIQDAD DIIQDAD (0909047) D1JUAND Py

"* QUUT'T D7JaqvS2 D}JaqvSt (DUOAIYISVG) DUNT

OST) Tena Ow eh to ren by oroInsnig snpiaty SNUOIOYINT
baa *yoreure'y Dwojso107Yy2 visnduvT

‘*(uryoin eas Jo soutds) -ds snjosjuasosajazy

Hise iene ouury v2.00 (szoxpoyuns) $2]01]DF7

“**QuUr'y wnjoursed (wniu0.lfoy) wniud.ifoy
x x Dt tale ass ners ouUr’y s2zuaduas jndva (vuosiavy) vidvsoLy

scar trae, QUUT'T Vjoajay (VIADSOL) DIUDSOLT
Sate SUUT'T DS0La S049 (DIADSOL) DIADSOLT

+ diapoig sunsaja snurns (vgniq) vdnaq

SUN eres ouUl'y x2psad (wniyoq) wnyoq

“ ' ouIey 2y24DUuv] DjvI1Gnp Drr4aWo0]gIG,

BY Nrasdlo shai ouur’y xud) xud] (vu2sudT) vaniqa’)

“75s 5+ QUUT'T DpDUIgoss (visvgos4vjnNIG) vUI]]a}0194/)

oe Gia ena suUl'y wndp.f (wndo47) wnjno4.0/)

"Tot" UdOg snavsqay snavagay (sijnaysung) snuo’y
Smale nna Ovian ehee.t teks, cakener ere spulyy pyowuwuas pany’)
Bipia amier tikes peuley (euxeriey (o ehu beset user ouury Dlagsv vav}’)
tts Sse aoInsnig wosojnpou wnryj213’)
** QUUT'T wnjnsaggiad (sngwo.soc) wunrsDuD’)
os se - ee Spee a Var

Persp e Laks ouury yoL0yap (siydvsp) siydosy
“* uayosnaypy vygvos vjonbyup (voap) vaLp

“yoreure’y wnpisoyf (snqwo.jsog) wnrswuvy

162

baa bss

eset,

“

ovld
uepeyD

eltyy
ueyeyy)

Te31@

TOYS
soy
nepne’y

V osnoyy
nepne'y

x KK MK

uefqo

iWeb tteia an en ors) Sate UUL'T 2jDjNII704 (27¢4]32427) snua 4
el eieiglotsia Baiatreiehs auury viadiand (07¢4)32427) snua 4
Ro egartang SoG: auUr'y] 07041290) (0)G«]G14aq) snuay
SPS NGA inp nis Bs cod sas auUT'T wnpjauigin) (unspy) wnsvy
teense UI[IUIL) SntiaasDgs (DUOJSOsOULAD Jy) Og4n J,
ie eens aUUT'T Snwopsosdiys (DUO}SosWAD Py) Ogan J,
ajlinvtalres i asi ei lal'eite na lel ees Bre ter/mudy eye. "b) aS (epnossdo) ouul'y

SNUOISOAMGID SNUOISOLAG1D (DUO]SOLOWAD YY) Og4nT,
ese: ouUr'y SnjojnovU snzDjnavU (snYyI0L J) SNYI0L [,
Toots ss yQTBUTeT Ssnppsspsqut (snyr04 J) SnYyI0L],
moiS Abs: be mya ‘slip eho) se) Siigiere ave ouur'y psowonbs Duzopis J,
weiter erpowee ke ouur'y oyvyngns (vigasa ].) 04G042 [.
srtrss sss yoreurey pppjynuas (Durpjaind py) vigasa J,
Ht eer en err tee ulog siupidd (unjoa 7) wnj20 J,
a) ohare 6a, 36,0 a Da sl ahaa ais zyuUIsY s7vonp snjqpuods
pce Oe een ee naire atie oe SESE et cae val wale at elie e tenet 4 ouury odin} SN]]OYN
exe Supe, weer scale auury sapeu (snuoz0z14y) snuor0z1yy
sto tst ss gpepory wopojod (sndvdiupimd) suvspon?y
po Sica te tac YOIRUIL'T Siquy) (0499043}q) 04199013}
situs, cata dlatsars, ASUS ulus viu0dag (0429040) 0499040}
Mee woes © oo seo eBUIBey 0rg409s (od0qu077) D199042} J
icvowcerit i ep ep oR ouur’y 2dvs1y9 (odvduvzy) 04990La1q
Troe | UTTQUUIL) DUO]SoUDjaU (D]jIUD Py) sanUIj0g
Perera ha yoreure’y snpranpjad (auowpoivg) 40}1g
Charan € Sie hs werd Mecloagcie et obs Te ieayre yoleue’y DD] jNINI Da4ISQC)
Let oh gn secNe ta cen searches guury ojrjog (Djtsauryguyp) DIAN
TRO STOR IE es QUUT'T DjaUu0W DjauoU DILDJaU0 PY

panuiuoj—SALIS GALVAVOXA WOU TIAHS GaAXYOMNN

163

164 MARIANAS PREHISTORY

A third bone point, slender and long (134 mm.) was also found at
Laulau House A, Level 1 (third specimen from top, fig. 88). This may be a
fish spear point. It is interesting to observe that it has the same form as
ancient Hawaiian arrow points; the ancient Hawaiians used bows and
arrows for sport, in shooting rats, though the bow and arrow were un-
known in the Marianas at the time of first contact.

A partially sawed section of human long bone was also obtained at Blue I
(lowest specimen, fig. 88).

ORNAMENTS

One fragment of Babirussa tusk is illustrated (fig. 86, lower right speci-
men). It is 82 mm. long, was a surface find at Chalan Piao, and may have
been used as an ornament.

UNWORKED ANIMAL BONES

Unworked animal bones, including remains of dogs and pigs, were
found in the excavations. The Chamorros at the time of discovery were
reported to be without either dogs or pigs, which is highly unusual for an
Oceanic people. Unfortunately, all the dog and pig finds listed below oc-
curred less than six inches below the surface, so the probability that these
may represent modern intrusions cannot be excluded.

Specimens and Provenience—Humerus, tibia, radius, fragment of scapula,
astragalus, and tooth of young pig (Objan, Level 1, Saipan); fragment of
pig tusk (Chalan Piao, surface, Saipan); femur of young carnivore, prob-
ably a dog (Objan, Level 1, Saipan); canine tooth of dog (Laulau Rock
Shelter, Level 1, Saipan); vertebra of seal (Blue I, Level 1, Tinian); three
fragments of turtle plastron (Objan, Level 4, Saipan). The bones were
identified by Dorothy B. Foss, formerly osteologist at Chicago Natural
History Museum.

UNWORKED FisH BONES

Most of the remains of fish found in the excavations were teeth, man-
dibles, and maxillas of parrot fish. According to Loren P. Woods, Curator
of Fishes at Chicago Natural History Museum, at least three species of
Callydon are represented, but they are so fragmentary that specific identifi-
cation is not possible. Parrot fish teeth may well have been used as cutting
and scratching tools. Provenience of parrot fish bones: Objan (XX’ profile,
Levels 2, 4, 5); Blue I (Levels 1-5).

Vertebrae of marlin (Makaira) were found at Objan (Level 1) and one
tip of the spear of what is probably Makaira was also found at Objan (Level
6). A fragment of the ramus of a sparid, probably Monotaxis, was found at
Laulau House A (Level 1). The hypural vertebrae of a scombid, a tuna, or

SHELL, BONE, AND METAL ARTIFACTS 165

Fic. 88. Upper two specimens: fragments of spear points cut from human bone.
Third specimen from top: bone fish spear point(?). Bottom specimen: partially sawed
section of human long bone. Length of top specimen, 133 mm.

possibly a young sailfish were found at Objan (XX’ profile). A tooth plate
of the spiny puffer Diodon, probably hispidus, was recovered from Level 6 at
Taga. In addition, a miscellaneous collection of fish vertebrae, too frag-
mentary to be identified, was found at virtually all the excavated sites. A
large deposit of fish scales was found at the very bottom of Test A, Blue I.

METAL ARTIFACTS

Four metal artifacts (fig. 89) were found at the following sites:
Laulau House A (Saipan): Below the rocks packed around the base of shaft
9, at a depth of 1.0 foot below present ground level, was found what ap-

Fic. 89. Metal artifacts. Top row, from left to right: iron spear point, iron nail,
fragment of iron or steel knife blade. Bottom: copper stud-like object. Length of
upper left specimen, 101 mm.

166

SHELL, BONE, AND METAL ARTIFACTS 167

pears to be an iron spear point (probably a fish spear) and the fragment of
an iron or steel knife blade (upper left and upper right, fig. 89). There was
no evidence to indicate that these specimens were intrusive and I believe
they are contemporaneous with the original occupation of Laulau House
A. If this is so, the site is historic and its date lies between the discovery of
the Marianas by Magellan in 1521 and the final Spanish conquest and
abandonment of Saipan in 1698.

Objan (Saipan): A heavily corroded fragment of a square, iron nail (top
center specimen, fig. 89) was found in Level 2, at a depth of 0.8 foot below
the present ground surface. The material above the nail was a dark gray
sand, loose and friable, and the possibility cannot be excluded that the nail
is intrusive and not contemporaneous with the site. On the other hand,
documentary evidence indicates that Objan was inhabited at the time of
Spanish conquest, and the iron nail may be of Spanish origin dating from
the early historic period.

Muchon Point (Rota): A copper, stud-like object (lowest specimen, fig. 89)
was found on the surface at the Muchon Point /atte. Its use and date are
unknown, though it is not recent. Local Chamorros had no idea what it
was originally used for.

VII. Conclusion

CULTURAL CHRONOLOGY IN THE MARIANAS

On the basis of the preceding description of excavations and artifacts,
what kind of a cultural chronology for the Marianas can be constructed?
The question demands a recapitulation of the evidence available on which
such a chronology can be based. Briefly, the evidence is as follows:

1. DocUMENTARY EVIDENCE

The early accounts of the first European explorers indicate that latte
type houses were in use at the time of discovery. Latte sites therefore as-
suredly go into the historic period. More particularly, Objan is mentioned
as a site on Saipan occupied at the time of Spanish conquest.

2. TRADE MATERIALS OF THE Historic PERIOD

Two iron artifacts were found at Laulau House A, a latte site on Saipan.
One iron nail was found at Objan, though, as indicated previously, this
may be a modern intrusion.

There is no indigenous source of metal in the Marianas. It is an interest-
ing question whether the prehistoric Chamorros knew of metal through
pre-European contact with the Philippines, Japan or China, but there is no
evidence that they did, except for their avidity to obtain iron from the early
European ships. Certainly they did not secure iron in any quantity until
the Spanish commenced sailing regularly from Mexico to the Philippines.
For this reason, I believe that Laulau House A on Saipan dates from the
early historic period.

3. RADIOCARBON DATES

Through the generous co-operation of W. F. Libby and the University
of Chicago, two carbon 14 dates were obtained: A.p. 845 + 145 from Blue
I on Tinian, and 1527 B.c. + 200 from Chalan Piao on Saipan. Blue I was
shallow and undoubtedly a single occupation site. Chalan Piao had a much
greater depth of cultural deposit and spans a greater time period. The car-
bon 14 date was obtained from an upper level of the site. The four feet of
cultural material lying below the oyster shell from which the date was
obtained must be considerably older.

168

CONCLUSION 169

4. STRATIGRAPHIC ANALYSIS

(a) Depth of Deposits at Archaeological Sites and Duration of Occupancy.—Of
the excavated sites, Laulau House A, Chalan Kija, and Blue I were shal-
low and undoubtedly single occupation sites. Oleai also can be considered
a single occupation site. Objan, Laulau Rock Shelter, Chalan Piao, and
the House of Taga had a greater depth of cultural deposit and covered a
greater period of occupancy. Both Objan and Taga were Jatte sites; yet the
depth of deposit below the stone Jatte was great enough to conclude that
they had had occupants prior to the Jatte builders. In the case of Objan,
Laulau Rock Shelter, Chalan Piao, and Taga, there is no means for deter-
mining the total duration of occupancy of each site.

(b) Site Structure—The only site yielding any clear stratigraphy of its
structure was Laulau Rock Shelter, which gave evidence of two different
periods, each showing a different predominant burial practice. The upper
level of this site was given to cremations, the lower primarily to extended
burials laid in sand-lined pits. In the case of /atte sites, it is also interesting
to note that the remains of older Jatte structures underlying more recent
latte were not found. All latte so far found in the Marianas have been dis-
covered only on the surface.

(c) Pottery Seriation—A clear-cut pottery seriation, based on relative fre-
quencies of Marianas Red and Marianas Plain, was obtained from Laulau
Rock Shelter and Chalan Piao. There is no doubt that Marianas Plain, the
predominant pottery type at latte sites, was preceded in time by Marianas
Red. Artifacts other than pottery yielded no significant seriation data.

On the basis of the kinds of evidence listed above, what is the chronologi-
cal relationship of the sites excavated? They are listed below, in their
admittedly somewhat tentative chronological order.

Chalan Piao (Saipan): The levels contemporaneous with and below (1.5
feet and below) the oyster shell yielding the carbon 14 date of 1527 B.c. +
200 are the oldest evidence of human occupation in the Marianas. These
levels also showed a predominance of Marianas Red sherds. The surface of
the site is undoubtedly much later, as the predominant pottery type on the
surface is Marianas Plain, characteristically associated with /atte sites.

Laulau Rock Shelter (Saipan): The presence of an appreciable number
of Marianas Red sherds in the lower levels indicates that the lower section
of this site is old. The upper level, consisting of the ashes from cremations,
is certainly much younger. In this upper level, no Marianas Red sherds
were found.

Taga (Tinian): Before the giant latte were constructed, the House of
Taga was occupied by people who were probably not Jatte builders. We
do not know how far back in time the lower levels go. Thompson (1932,

170 MARIANAS PREHISTORY

p. 30) notes that Hornbostel found sherds of Marianas Red at the House of
Taga, indicating that part of the site may be contemporaneous with the
lower part of Laulau Rock Shelter. We do not know when the great shafts
and capstones were erected on the surface. They obviously, together with
As Nieves on Rota, represent the climax of Jatte building, and a develop-
mental period probably preceded them. My guess is that the Taga and As
Nieves latte are roughly contemporaneous and that they date from a.p. 800
to 1400, but this is pure conjecture.

Oleai (Saipan): At this non-/atte site a few sherds of Marianas Red were
found, though the dominant pottery type was Marianas Plain. In the
seriation of Marianas Plain rim types, Oleai falls later than Taga, though
earlier than Blue I. Otherwise, there is no evidence on which to base a
date.

Objan (Saipan): Like Taga, the lower levels of Objan may precede the
latte-building period by a considerable margin. Though I have not noted
the fact previously, Marianas Red sherds were found on the bulldozed part
of the site, so at least they are present, even if the stratigraphic evidence has
been destroyed. The Objan /atte, however, no doubt go into the historic
period, as the village is mentioned in seventeenth century Spanish ac-
counts.

Blue I (Tinian): This shallow, single-occupation Jatte site yielded a car-
bon 14 date of a.p. 845 + 145. The significance of this date is that it is the
first evidence found of the duration of the Jatte-building period in the
Marianas.

Chalan Kija (Saipan): This single-occupation Jatte site, yielding Mari-
anas Plain but no Marianas Red, is certainly not particularly early. In the
seriation of Marianas Plain rim types, it falls with Oleai. Otherwise there is
no evidence on which to base a tentative date.

Laulau House A (Saipan): All the evidence—pottery, /atte, and presence
of two iron implements—points to a late date for this site. I believe it dates
from the historic period, probably between 1600 and 1698.

The chronological results of the Museum’s archaeological work in the
Marianas can be summarized as follows: On the assumption of the reliabil-
ity of the Chalan Piao carbon 14 date, the Marianas were settled by 1527
B.c. + 200. These first settlers were probably fishermen and farmers, just
as the Chamorros were at the time of European discovery of the islands.
At a much later date, a new style of house construction—the building of
/atte—was introduced, either as an indigenous development or as the result
of outside influence. At the same time, Marianas Plain pottery superseded

| Marianas Red. The earliest date from the Jatte-building period is the A.p.

845 + 145 date from Blue I. Latte-building, and the characteristic artifacts

CONCLUSION tra

that went with it, continued up to the historic period and the abandon-
ment of Saipan in 1698.

The framework of this chronology rests on only two radiocarbon dates.
It would be a considerable comfort to have more. Until additional dates
are obtained, either from the Marianas, or from the other two major
Western Micronesian island groups—Yap and Palau—these dates must
suffice for the construction of working hypotheses; they provide a valuable
clue to the nature of the culture history of Micronesia.

LATTE PHASE OF MARIANAS CULTURE

A principal aim of the Museum’s work in the Marianas was to clarify
what I have termed the Latte Phase of Marianas prehistory. The results of
the field research have furthered this aim in two ways: (1) Through the
carbon 14 date of a.p. 845 + 145 obtained from Blue I some insight has
been obtained as to the time span of this phase. On the basis of this date,
the Latte Phase endured for at least 850 years, up to the historic period.
(2) Through the survey and excavations, a clearer picture has emerged of
the culture content of this phase. With few exceptions, the artifacts de-
scribed in the preceding pages refer to the Latte Phase and provide, when
combined with the archaeological and ethno-historic data previously pre-
sented by Fritz (1904), Safford (1905), and Thompson (1932, 1945), a pic-
ture of prehistoric Chamorro culture, which apparently remained rela-
tively stable up to the time of the Spanish conquest.

To itemize the artifact content of the Latte Phase would merely result in
a listing of what has already been described. It may be useful, however, to
outline certain of the characteristics of the Latte Phase, as revealed pri-
marily by archaeological work.

The Village: The largest concentration of population was the village,
which apparently varied considerably in size. The problem of size is one of
determining the number of houses in a single village. Thompson has rightly
noted that in the case of many villages all the houses were not of the stone
latte type but were built on wood houseposts, which have long since dis-
integrated. We do have records of three villages, where apparently the
majority of houses were of the stone latte type. Two were briefly described
by Hornbostel and consist of Agingan on Saipan (completely destroyed in
World War II) and Taga on Tinian (Thompson, 1932, pp. 18-19). The
third—the Blue Site on Tinian—has been described in this report. Taga
consisted of 18 stone latte, Agingan of 12 /atte, and the Blue Site of 10 Jatte.
I believe the houses were of sufficient size to accommodate an extended
family group. If we assume an average of 15 people per house, and further
assume that the houses were all occupied at the same time, this would give

172 MARIANAS PREHISTORY

a population of 270 people to Taga, 180 to Agingan, and 150 to Blue I.
I believe these are minimal figures, as it is highly likely that additional
houses of wood posts only were constructed at each village, judging from
the sherd areas surrounding Blue I. At the same time, the village units were
apparently never very large.

The size of villages must also be matched against population estimates
for the Marianas as a whole. Sanvitores, the head of the Spanish mission
when it was established in 1668, estimated the total Chamorro population
at approximately 100,000. I have previously suggested 50,000 as a more
realistic guess (Spoehr, 1954, p. 56).

An interesting feature of the Taga, Agingan, and Blue sites is the oc-
currence of the largest /atte as the central structure of the site. Ethnological
evidence indicates the presence of men’s houses and a chiefly class among
the Chamorros at the time of contact. The oversize central latte may repre-
sent men’s houses or the residences of chiefs. I do not believe they were re-
ligious structures, reserved only for religious purposes, similar to the hezau
of Hawaii or the marae of Tahiti, because the Chamorro central /atte had all
the debris of daily living around them, and because there is no ethno-
historic evidence of true temples reserved only for religious rites in the
ethnological literature on the Marianas.

The location of latte type villages was evidently controlled by accessibil-
ity to available resources. Like other island peoples in Micronesia, the
peoples of the Marianas relied on the sea as the principal source of protein.
It is therefore not surprising that latte villages are primarily located along
the shore, in places that combined adequate farming areas with easy access
to the sea. Though some interior Jatte sites have been found on Tinian,
Saipan, and Guam, they are still not so far removed from the sea as to
eliminate it as a food resource.

Houses: All latte have around them the debris of daily life. In the Mu-
seum’s excavations, certain features of living associated with household ac-
tivities were discovered. At Objan and Laulau House A, the remains of
work areas paved with coral pebbles were found around the stone shafts
and on the seaward side, indicating the presence of work areas under the
raised floor of the house. At the rear of Laulau House A, two earth ovens
were uncovered, demonstrating that the ancient Chamorros made use of
this widespread Oceanic manner of cooking. Though no evidence of sepa-
rate cook houses was found, their presence on modern Saipan and their
wide distribution in Micronesia today make it probable that each latte
house had a frame cook house adjacent to it.

I have said elsewhere that the latte may well have housed extended fam-
ily groups. The evidence for this is the size of the structures, relatively large

CONCLUSION 173

for Micronesian houses, and the rather wide occurrence of extended fam-
ilies as residence groups in Micronesia.

Technology: The latte-builders were prolific if not expert potters. Mari-
anas Plain is the dominant pottery type, associated with much lower fre-
quencies of Marianas Fine-Line Incised, Marianas Cord-Marked, and
Marianas Trailed. Marianas Red, and the lime-impressed sherds occurring
principally at Laulau Rock Shelter, pre-date the Latte Phase.

For cutting tools, the /atte-builders relied on shell and stone, with the
former predominating. Evidence of agriculture is limited to finds of mor-
tars and pestles, together with frequent, if inadequate, references to Cha-
morro farming in the ethno-historic literature. The latter indicates the cul-
tivation of rice, though it may have been millet. As the Latte Phase goes
into the historic period, I believe it highly probable that the cultivation of
rice was a characteristic of the Latte Phase as a whole. Although remains
of both dogs and pigs were found, they occurred as shallow finds; they sug-
gest, but do not prove, the association of these animals with the latte people.

Fish hooks and gorges are characterized by a paucity .of forms, particu-
larly compared to Polynesia. No archaeological evidence of canoes was
found, though they must have been made and used. Remains of seal and
turtle suggest that both these animals were used as food.

We have no archaeological evidence of clothing. Pandanus mat impres-
sions on sherds are the only evidence of the textile art.

War: Barbed spear points of human bone and an abundance of sling-
stones testify to the use of these weapons in war. Ethno-historic data reveal
the ancient Chamorros as a relatively war-like people.

Ornament and Art: Ornaments found were confined to a few simple
beads, bracelets, and pendants. No carving was uncovered. It is probable
that Latte Phase art was expressed in perishable materials that have long
since disappeared.

Disposal of the Dead and the Evidence of Burials: Both extended and second-
ary burials (consisting of miscellaneous collections of bones) were found
associated with Jatte sites. Also, it is possible that the cremations forming the
upper level of Laulau Rock Shelter are associated with the Laulau latte. A
variety of burial practices therefore was common. Grave goods were ab-
sent, except for what were probably trophy jaws. Some of the burials at
Objan lacked the skulls, which must have been removed prior to burial.
Brown-stained teeth of some burials are evidence for the antiquity of betel-
nut chewing. One skull and several long bones found at Blue I exhibited
pathology suggestive of yaws (Stewart and Spoehr, 1952).

174 MARIANAS PREHISTORY

PRE-LATTE PHASE CULTURE

Evidence as to what came before the Latte Phase is confined primarily to
a distinctive pottery type, Marianas Red. Other evidence expressed in arti-
facts consists merely of a few stone and shell adzes, not of substantial use,
since they are insufficient statistically or not of distinctive form. The ex-
tended burials lying in sand-lined pits, found in the lowest levels of Laulau
Rock Shelter, complete the list. On this scanty evidence, nothing as pre-
tentious as a “‘culture”’ can be inferred and delineated. On the other hand,
Marianas Red and the lime-filled, impressed pottery from Laulau Rock
Shelter do provide distinctive pottery types that should in the future be of
great aid in establishing culture relations with other islands and in provid-
ing material for cross-dating with other islands.

In this connection, it seems quite unlikely that truly major culture
change will ever be delineated in the prehistory of either Micronesia or
Polynesia. The first settlers of these islands undoubtedly brought with
them the basic adaptations in fishing and agriculture that allowed them to
survive and that persisted as the means of making a living up to historic
times. The islands limited large-scale concentrations of population, acting
as points of diffusion and cultural interaction over a major region, as con-
trasted with continental foci of culture growth. On the other hand, the
probably narrow range of cultural variation through time in Micronesia
and Polynesia, and the basic similarities so far uncovered in archaeological
and historic cultures, make ethnographic facts essential in archaeological
study. In Oceania, archaeology and historical ethnology are closely joined.

CULTURAL AFFILIATIONS OF THE MARIANAS

Until archaeology is further advanced in the western Pacific, it is too
early to determine the prehistoric cultural affiliations of the Marianas.
Logically, on the prehistoric time level, the Marianas should exhibit rela-
tions with the Palaus and with Yap, probably also with the central Caro-
lines. But until a considerable amount of archaeological work has been
done in these areas, no determination can be made.

Even more important is the nature of relations with the Philippines. The
linguistic evidence indicates that both Chamorro and Palauan fall with the
Philippine languages rather than with those of the rest of Micronesia
(Grace, 1955). The presence of red pottery wares in the Philippines may
link with Marianas Red, and I believe that through these red wares a rela-
tion will be established. Ethnologically reported similarities such as the cul-
tivation of rice, as well as minor traits such as lining dug-out pits with
sharpened stakes to catch unwary enemies, point to cultural connections.
Here again, until stratigraphically established sequences and careful de-

CONCLUSION 175

scriptions of prehistoric pottery and artifact types are available in the liter-
ature on Philippine archaeology, the cultural affiliations of the Marianas
with the Philippines must remain an open problem. I have little doubt
that the Marianas are closely related to prehistoric Philippine cultures, but
this is a personal impression and cannot be substantiated by published
evidence.

REFLECTIONS ON OCEANIC PREHISTORY

The remainder of this report is speculative in nature and deals with the
ways in which future archaeological work in Oceania can proceed in most
productive fashion. In large part, the discussion that follows is a rephrasing
of an earlier article (Spoehr, 1952) and deals with these questions: (a)
What is the most useful spatial framework within which to view the prob-
lems of Oceanic prehistory? (b) On the basis of archaeological work, what
actual chronology has been achieved? (c) What is the strategy that should
govern archaeological field work in Oceania? (d) What lines of research
allied to archaeology promise the most fruitful results for Oceanic culture
history? ;

SPATIAL FRAMEWORK OF OCEANIC PREHISTORY

It is characteristic of areas in which archaeological work has long been
pursued that there is a constant revaluation of the spatial framework in
which problems of culture history are set. As soon as a respectable chronol-
ogy has been achieved and some insight gained into the specific character-
istics of cultural development within a given area, the spatial framework
for the interpretation of that area’s cultural history comes under critical
examination and is altered and refined. In New World archaeology, one
need only point to the manner in which archaeologists working in the
American Southwest have gradually reformulated their working concepts
as to the principal areal subdivisions of the Southwest, as these shift through
time. Other examples are the formulation of the concept of ‘‘Meso-
America” as providing a principal division for the interpretation of Middle
American culture history (Kirchhoff, 1943), and current interest in ‘“‘Nu-
clear America”’ as a larger area accommodating a major continental cul-
ture growth (Kroeber, 1948; Willey, 1955).

In Oceania, a thorough evaluation of the spatial framework in which to
set the problems of Oceanic culture history has not been made. Through
long usage, the division of island Oceania into Polynesia, Micronesia, and
Melanesia has become established as the most convenient classification of
its peoples and cultures. For purposes of ethnographic description, this
classification has been useful and will no doubt retain its essential validity.

176 MARIANAS PREHISTORY

When the tripartite division of the islands is imposed on what is known
of their culture history, however, I have serious doubts as to the utility of
the present classification. Because it is so universally accepted, the tradi-
tional areal division can easily obscure rather than clarify the nature of
prehistoric events in Oceania. Its utility is in need of review and evalua-
tion.

Although an adequate reconstruction of Oceanic prehistory cannot yet
be made, at least two phases of its past are apparent. The first is the early
settlement of Australia, Tasmania, at least parts of New Guinea, and pos-
sibly the Bismarck Archipelago. A conservative view places the peopling of
the New Guinea-Australia region during the fourth glacial. ‘The settlement
of this area was affected by a series of paleogeographic factors, reviewed in
interesting and competent fashion by Keesing (1950). On the basis of the
cultures of Australia and what can be surmised of early Negrito culture,
these first settlers possessed only the simplest form of water transport, and
were food collectors rather than agriculturalists.

A second chapter in Oceanic prehistory occurred at a much later date.
This was the settling of the islands of Micronesia and Polynesia, which
have always been separated by great expanses of ocean. For human migra-
tion into these islands, two conditions had to be met. The first was the
invention and development of a form of sea-going craft, and associated
techniques of seamanship and navigation, which allowed man to traverse
truly awe-inspiring distances across the open sea. The invention of the sail-
ing, single outrigger canoe and the Polynesian double canoe fulfilled this
requirement. The second condition was the domestication of those food
plants particularly characteristic of the Indo-Pacific region, especially
taro, yams, the banana, and breadfruit. With these conditions met, the
stage was set for the movement of Malayo-Polynesian-speaking peoples
across the Pacific in one of the greatest periods of exploration and settle-
ment in human history. It is true, of course, that we do not know whether
the first of these settlers spoke Malayo-Polynesian languages. Yet the fact
that Micronesians and Polynesians all belong to this linguistic family
makes it probable that the first settlers did also.

It is at this point that the problem of an adequate spatial framework for
Oceanic prehistory becomes important. Some new areal concept is neces-
sary to accommodate the events associated with the spread of canoe-
voyaging, probably Malayo-Polynesian-speaking, peoples across the Pa-
cific. The division of island Oceania into Polynesia, Micronesia, and Mela-
nesia tends to obscure our perspective on this movement. It was a move-
ment participated in by both Polynesians and Micronesians, who share
significant sectors of their culture.

CONCLUSION sya

In a previous paper (Spoehr, 1952), I proposed the term Micro-Polynesia
as an areal concept that would embrace the movement of Malayo-Poly-
nesian-speaking peoples into the Pacific islands. The difficulty with this
term is that it does not do justice to the spread of Malayo-Polynesian canoe
voyagers into Melanesia. Current linguistic work (Grace, 1955) indicates
an early Malayo-Polynesian influence of wide extent in Melanesia, and
any areal term should take this into account. Some new spatial concept
should be devised to accommodate what seems to me to be a distinctive
chapter in Oceanic prehistory.

CHRONOLOGY

Within the area encompassed by the spread of canoe-voyaging, Malayo-
Polynesian-speaking peoples, what absolute chronology has archaeological
work produced? Since World War II, the beginning of a framework of
radiocarbon dates has been achieved. At the time of writing of this report,
the following radiocarbon dates are available:

1. Oahu, Hawaii: a.p. 1004 + 180. Excavations by Kenneth P. Emory. Radiocar-

bon date determined by W. F. Libby, University of Chicago (Libby, 1951).

2. Tinian, Marianas: a.p. 845 + 145. Excavations by Alexander Spoehr. Radio-
carbon date determined by W. F. Libby, University of Chicago (Libby, 1952).

3. Saipan, Marianas: 1527 B.c. + 200. Excavations by Alexander Spoehr. Radio-
carbon date determined by W. F. Libby, University of Chicago (Libby, 1952).

4. Viti Levu, Fiji: 46 B.c. + 500; a.v. 654 + 500; a.v. 754 + 500; a.v. 1002 + 300;
A.D. 1255 + 300; a.p. 1305 + 300. Excavations by E. W. Gifford. Radiocarbon dates
determined by H. R. Crane, University of Michigan (Gifford, 1955).

5. New Caledonia: 847 B.c. + 350; 482 B.c. + 400; a.v. 73 + 350; a.p. 253 + 300;
A.D. 618 + 300; a.pv. 1048 + 300; a.p. 1168 + 300; a.v. 1338 + 300; a.v. 1568 + 300.
Excavations by E. W. Gifford. Radiocarbon dates determined by H. R. Crane, Uni-
versity of Michigan (Gifford, personal communication).

These dates will no doubt soon be supplemented by additional ones
from Hawaii and New Zealand, where systematic archaeology is proceed-
ing apace. Additional dates are also badly needed for western Micronesia.
Although the dates presently available are hardly sufficient to establish a
firm chronology for the spread of Malayo-Polynesian peoples into the
Pacific, they do provide a basis for working hypotheses which may be
altered or substantiated as new evidence accumulates. Two such hypothe-
ses are recapitulated below from a previous paper (Spoehr, 1952).

The first hypothesis is that by approximately 1500-2000 B.c., man,
equipped with sea-going canoes and the plants necessary for his survival in
an Oceanic island environment, had commenced his eastward migration
from the general Malaysian region.

1 Since this was written, the Fiji and New Caledonia dates have been fully reviewed
in Gifford and Shutler (1956).

178 MARIANAS PREHISTORY

The second hypothesis is that the eastward migration through Micro-
nesia and Polynesia spanned approximately 2500 years. Terminal points of
this period are set by the Saipan and Oahu dates, the Hawaiian Islands
certainly being one of the last regions to be settled in Polynesia. However,
current archaeological field work by Emory in Hawaii suggests that the
first settlement of the Hawaiian Islands may be several hundred years
earlier than the single radiocarbon date now available.

The dates obtained by Crane from Gifford’s excavations in Fiji and
New Caledonia are also an extremely important part of the picture. It is
my considered guess that Fiji was settled by 500 B.c. If this is so, nearby
Samoa and Tonga in western Polynesia were probably populated shortly
thereafter. Emory agrees that the nuclear center of eastern Polynesia—the
Society Islands—was settled about the beginning of the Christian era.
New Zealand may be somewhat earlier than Hawaii, but a series of dates
from each of these two areas must be determined before an answer can be
forthcoming.

The 847 B.c. date from New Caledonia is of exceptional interest. The
linguistic evidence suggests an early Malayo-Polynesian influence in Mela-
nesia, and the radiocarbon date confirms this supposition. My guess is that
one of the earliest movements of Malayo-Polynesian canoe voyagers was
along the northern coast of New Guinea and into the Bismarck Archi-
pelago, that they interbred with peoples already resident there, and that
they pushed southward along the Solomons, the New Hebrides, and finally
to New Caledonia, the southernmost major island in Melanesia.

Within the framework of the tentative chronology posed above there are
many unanswered questions. What was the nature of these sea-voyagers’
expansion eastward? Did it occur as a slow trickle or as a series of bursts or
waves of exploration and settlement? What of the still unknown paths of
migrations? In a previous paper (Spoehr, 1952), I committed the indiscre-
tion of stating that the Polynesians probably came through Micronesia.
This view is not sustained by convincing evidence and the question must
remain open.

Most important of all is the nature of culture change encompassed by
the chronology derived from our present carbon 14 dates. Here the answer
can only come from careful excavations in key island areas. Since World
War II, an excellent beginning has been made, yet much remains to be
done.

STRATEGY OF ARCHAEOLOGICAL WorRK

Except where anthropology is professionally established, as in New Zea-
land and Hawaii, islands deserving careful archaeological work are far
removed from institutional centers of research, and access to such islands is

CONCLUSION 179

sometimes difficult and arduous. Archaeological field work is therefore ex-
pensive, and if for no other reason, the strategy of field work assumes great
importance.

Two papers on the status and problems of archaeology in Micronesia
and Polynesia should be noted. One (National Research Council, Sub-
Committee on Pacific Archaeology, 1951) deals with Micronesia and points
particularly to the critical importance of the Palaus and Yap in the pottery
area of western Micronesia. Since the publication of this article, Osborne
has conducted survey and excavations in the Palaus, and Gifford has
worked on Yap. The results of their efforts should contribute greatly to
archaeological knowledge of Micronesia.

In the Marianas, the gaps in this present report can best be filled by
excavations on Rota. Of all the islands in the Marianas, the greatest num-
ber of sites remain on Rota, which has suffered less destruction of archaeo-
logical remains than Guam, Tinian, or Saipan. What is particularly
needed is field research to determine the chronology and culture content of
the pre-/atte period.

Also of significance are Truk, Ponape, and Kusaie, strategically placed
across the hundreds of miles of the Caroline chain. Thorough investigation
of the impressive stonework sites of Ponape and Kusaie would in itself be an
exciting task.

The second paper bearing on the strategy of archaeological work is
Emory’s excellent statement on the needs of Polynesian archaeology
(Emory, 1953). Fortunately, there is a body of descriptive data on surface
structures and artifacts in Polynesia. Added to this is the very promising
current work, conducted with due regard for stratigraphy, now taking
place in New Zealand and Hawaii, and of which Duff’s report on the moa-
hunter period in New Zealand is a most productive result (Duff, 1950). It
should not be long before the outlines of prehistory in both Hawaii and
New Zealand become part of the published record.

For other considerations affecting Polynesian archaeology, the reader is
referred to Emory’s paper, though I cannot let the opportunity pass with-
out seconding Emory’s stress on the need for careful excavations in the
Society Islands, Samoa, and Tonga.

The Melanesian picture is much more complex. Gifford’s work in Fiji
(Gifford, 1951) and in New Caledonia is a real contribution. The next
logical step is perhaps an extension of the line of attack through the Solo-
mons and into the Bismarck Archipelago, with one goal the possibility of
defining the area of Malayo-Polynesian influence in prehistoric times in
Melanesia. Melanesia as a whole, however, encompasses such a great area,
its prehistory is undoubtedly so complex, the relevant time scale covers so

180 MARIANAS PREHISTORY

many millennia, that any planned program of archaeological research in
the area presents almost overwhelming difficulties. The prehistory of New
Guinea alone would take decades of work to unravel.

Consideration of Malaysia is beyond the scope of this report, except to
comment that determining the relationships of the Oceanic islands and
Malaysia is an ultimate goal of prehistoric research in the Pacific. Heyer-
dahl’s courageous voyage from Peru to Polynesia and his subsequent mono-
graph have excited general interest in the question of trans-Pacific contact
with the New World. That contact took place I have little doubt, but a
balanced view of the evidence does not show that it had major effect on the
course of culture history in the Pacific. The prehistory of Oceania is an off-
shoot of the Old World, and in time will be shown in its proper relation to
the ancient culture growths of Malaysia and the Asiatic mainland.

PRoMISING LINES OF RESEARCH ALLIED TO ARCHAEOLOGY

As archaeological research proceeds in the Pacific, there are within the
general field of anthropology two allied lines of investigation which have
already contributed substantially to Pacific prehistory and whose further
pursuit promises most significant results. These allied lines of investigation
are linguistics and physical anthropology.

Although I am not a specialist in the field of linguistics, and hence not
perhaps in a position to offer a considered opinion, the development of
structural linguistics in the last few decades appears as a major advance.
From the point of view of Pacific culture history, an improved understand-
ing of the historical relationships of Pacific languages, particularly those
included in Malayo-Polynesian, is an absolute necessity. ‘Through the ap-
plication of the analytical techniques of structural linguistics to compara-
tive studies of Pacific languages, the ultimate historical synthesis will have
much greater validity, on both phonemic and morphological levels, than
it ever could have had thirty years ago. At the same time, although glotto-
chronology seems still to be the subject of considerable controversy, it
promises to be a valuable technique, particularly when combined with
radiocarbon dating, for determining chronology and historical relation-
ships. In a recent paper, Elbert (1953) has applied the glotto-chronological
method to the Polynesian languages. The present research of the Tri-Insti-
tutional Pacific Program, sponsored by Yale University, the University of
Hawaii, and Bishop Museum, has as one of its first objectives the gaining of
a clearer understanding of the historical relationships of the Malayo-
Polynesian languages of Oceania, through the application of the newly de-
veloped linguistic techniques. To this end, Grace (1955) has recently pub-
lished a tentative classification of the Malayo-Polynesian languages of
Polynesia, Micronesia, and Melanesia. Under the Tri-Institutional Pacific

CONCLUSION 181

Program, he is now testing this classification through extensive field work,
primarily in Melanesia, whose linguistic complexity makes it of critical
importance.

In physical anthropology, the extension of research on human blood
groups to the indigenous peoples of the Pacific certainly marks the most
significant advance made in knowledge of the genetic make-up of Pacific
populations. The spatial limits of the present distribution of Malayo-Poly-
nesian-speaking peoples encompass genetically diverse groups. It is an
interesting question whether the original movement of canoe voyagers
eastward from Malaysia was undertaken by peoples of greater or lesser
homogeneity, from either the point of view of morphology or of blood
group frequencies, than modern Polynesians, Micronesians, and Malayo-
Polynesian-speaking Melanesians. It is possible that Melanesia may be
somewhat similar to Madagascar, where Malayo-Polynesian-speaking
migrants have been merged into a largely African population and have
lost their original identity, although the language brought by the migrants
has remained dominant.

A final comment concerns another line of investigation that is only re-
cently receiving the recognition it deserves. Particularly in Micronesia and
Polynesia, man moved into the islands, bringing with him a distinctive
biotic assemblage of economic plants, weeds, domestic animals, and organ-
isms parasitic on man himself. The evolution and history, and the manifold
ecological relationships of these man-transported forms of life with their
human carriers, present challenging problems significant for an under-
standing of culture history (cf. Emory, 1953). Of prime importance are the
economic plants. The time is ripe for their intensive study, utilizing the
combined approaches of genetics, cytology, taxonomy, and plant physiol-
ogy in a manner comparable to the studies of New World food plants, par-
ticularly maize. Sauer’s intriguing hypothesis of an ancient center of do-
mestication in Southeast Asia of the principal starch-yielding cultigens of
the Pacific (Sauer, 1952), and Merrill’s stimulating, if often vitriolic, dis-
cussion of the dispersal of Pacific food plants (Merrill, 1954) are recent
publications from the pens of an eminent geographer and botanist, respec-
tively, that emphasize the need for a new program of research. However, it
is a program that demands the skills of the botanist, however valuable the
collaboration of the geographer and the anthropologist may be.

Significant also are competent zoological studies of the domestic ani-
mals—the pig, fowl, and dog—and that perennial hanger-on, the rat. And
lastly there are those organisms parasitic on man, of which the two most
important are probably the treponemas of yaws and filaria. In the case of
the former, some assistance may be rendered by paleopathology, granted

182 MARIANAS PREHISTORY

the difficulties in the diagnosis of pathological skeletal material (Stewart
and Spoehr, 1952).

What is called for, therefore, is a well-defined approach stemming from
what is not a well-defined field—human ecology. Yet because the princi-
pal objects of study are tangible life forms clearly related to man, an eco-
logical point of view is essential to progress in the understanding of culture
history in the Pacific.

Bibliography

Anson, GEORGE (Lord)

1748. A voyage round the world in the years 1740-44. John and Paul Knapton.
London. Compiled by Richard Walter.

ARAGO, J.

1823. Narrative of a voyage round the world in the Uranie and Physicienne corvettes
commanded by Captain Freycinet during the years 1817, 1818, 1819, and 1820.
Treuttel and Wurtz, Treuttel, jun. and Richter. London.

Beyer, H. OTLEY

1948. Philippine and east Asian archaeology. National Research Council of the
Philippines, Bulletin, no. 29.

Bowers, NEAL M.
1954. Problems of resettlement of Saipan, Tinian, and Rota. Honolulu.

Bryan, E. H., Jr.

1946. A geographic summary of Micronesia and notes on the climate of Micronesia.
U.S. Commercial Company. Honolulu. Typescript.

Buck, PETER H., Emory, KENNETH P., SKINNER, H. D., and Sroxgs, J. F. G.

1930. Terminology for ground stone cutting-implements in Polynesia. Polynesian
Society, Journal, vol. 39, no. 2.

Durr, ROGER

1950. The moa-hunter period of Maori culture. Canterbury Museum Bulletin,
no. 1. Wellington.

ELBERT, S. H.

1953. Internal relationships of Polynesian languages and dialects. Southwestern
Journal of Anthropology, vol. 9, no. 2.

Evsert, S. H., and SPo—EHR, ALEXANDER
1954. Linguistic studies in Polynesia. Nature, vol. 174.

Emory, KENNETH P.

1953. A program for Polynesian archaeology. American Anthropologist, vol. 55,
NOD:

Fritz, GEORG
1904. Die Chamorro. Ethnologisches Notizblatt, vol. III, no. 3. Berlin.

GiFForD, Epwarp W.

1951. Archaeological excavations in Fiji. University of California, Anthropologi-
cal Records, vol. 13, no. 3.

1955. Six Fijian radiocarbon dates. Polynesian Society, Journal, vol. 64, no. 2.

Grirrorp, E. W., and SHuTLER, Dick, JR.

1956. Archaeological excavations in New Caledonia. University of California,
Anthropological Records, vol. 18, no. 1.

183

184 MARIANAS PREHISTORY

GRACE, GEORGE W.
1955. Subgrouping of Malayo-Polynesian: A report of tentative findings. Ameri-
can Anthropologist, vol. 57, no. 2.

Hunt, Epwarp E., JR.

1950. A view of somatology and serology in Micronesia. American Journal of
Physical Anthropology, vol. 8, no. 2.

JosepH, AticE, and Murray, VERONICA F.

1951. Chamorros and Carolinians of Saipan. Harvard University Press. Cam-
bridge, Massachusetts.

KEEsING, FELIx M.
1950. Some notes on early migrations in the southwest Pacific area. Southwestern
Journal of Anthropology, vol. 6, no. 2.

KircHHorFrF, PAUL

1943. Mesoamérica: sus limites geograficos, composicién étnica y caracteres cul-
turales. Acta Americana, vol. 1.

KROEBER, ALFRED L.
1948. Anthropology. New York.

Le GosBIEN, CHARLES
1700. Histoire des isles Marianes, nouvellement converties a la religion Chrestienne,
et de la mort glorieuse des premiér missionaires qui y ont préché la foy. Paris.

Lipsy, W. F.
1951. Radiocarbon dating. University of Chicago Press. Chicago.
1952. Chicago radiocarbon dates, III. Science, vol. 116, no. 3025.

Marcue, A.

1891. Rapport general sur une mission aux Isles Mariannes. Nouvelles Archives
des Missions Scientifiques et Litteraires, vol. 1. Paris.

MERRILL, ELMER D.

1954. The botany of Cook’s voyages and its unexpected significance in relation to
anthropology, biogeography and history. Chronica Botanica, vol. 14, nos. 5-6.

MortTIMER, GEORGE

1791. Observations and remarks made during a voyage... in the brig Mercury.
London.

NATIONAL RESEARCH CouncIL, SUB-COMMITTEE ON PAciFIcC ARCHAEOLOGY

1951. A program for Micronesian archaeology. American Anthropologist, vol. 53,
no. 4

REED, Erik K.

1952. General report on the archaeology and history of Guam. National Park
Service, United States Department of the Interior. Washington, D.C. Typescript.

SAFFORD, WILLIAM E,

1905. The useful plants of Guam, with an introductory account of the physical fea-
tures and natural history of the island, of the character and history of its people,
and of their agriculture. Smithsonian Institution, Contributions from the United
States National Herbarium, vol. IX. Washington, D.C.

SANCHEz y Zayas, E.
1865-66. The Marianas Islands. Nautical Magazine, vols. 34, 35.

BIBLIOGRAPHY 185

SANGER, Rutu, Watsu, R. J., and Kay, M. Patricia
1951. Blood types of natives of Australia and New Guinea. American Journal of
Physical Anthropology, vol. 9, no. 1.

SAvER, Cart O.
1952. Agricultural origins and dispersals. American Geographical Society. New
York.

Stmmons, R. T., Graypon, J. S., SempLe, N. M., BrrpsELL, JosEPH B., MILBouRNE,
Joun D., and Leg, J. R.
1952. Acollaborative genetical survey in the Marshall Islands. American Journal
of Physical Anthropology, vol. 10, no. 1.

SoLHEm™, WILHELM G.
1951. Oceanian pottery manufacture. University of Manila, Journal of East
Asiatic Studies, vol. 1, no. 2.
1952. Paddle decoration of pottery. University of Manila, Journal of East Asiatic
Studies, vol. 2, no. 1.

SPoEHR, ALEXANDER
1952. Time perspective in Micronesia and Polynesia. Southwestern Journal of
Anthropology, vol. 9.
1954. Saipan. The ethnology of a war-devastated island. Fieldiana: Anthropol-
ogy, vol. 41. :

Stewart, T. DALE, and SPoEHR, ALEXANDER
1952. Evidence on the paleopathology of yaws. Bulletin of the History of Medi-
cine, vol. 26, no. 6.

THompson, LAURA

1932. Archaeology of the Marianas Islands. Bernice P. Bishop Museum, Bulletin
100, Honolulu.

1940. The function of latte in the Marianas. Polynesian Society, Journal, vol. 49.

1945. The native culture of the Marianas Islands. Bernice P. Bishop Museum,
Bulletin 185. Honolulu.

1947. Guam and its people. Revised 3rd ed. Princeton University Press-
Princeton, New Jersey.

WILLEY, Gorpon R.
1955. The prehistoric civilizations of nuclear America. American Anthropologist,
vol. 57, no. 3.

Index

Adzes, shell, 151-154; stone, 131-136

Animal bones, from excavations, 164

Anson, Lord George, 86

Arago, J., 86

Archaeological sites, destruction of in
Marianas, 28-29, 68, 73

Art, 173

Awls, 160

Beads, 158

Bernice P. Bishop Museum, 8—9, 180

Betel nut, prehistoric chewing of, 42, 157,
173

Blood groups, 181

Bowers, Neal M., 68

Bracelet, 160

Burials, 38, 40-43, 46, 54-57, 83-84, 96,
173-174

Cloud, Preston E., 8, 66
Cook houses, 45, 172
Copper, artifact, 167
Craneshe Re 7. 178

Dog, prehistoric, 25, 164, 181
Duff, Roger, 17, 179

Earth ovens, 46, 172

Elbert, S. H., 180

Emory, Kenneth P., 17, 137; 177; 178,
179, 181

Excavation methods, 27, 35, 45, 52, 58-
59, 60, 65, 78, 94-95, 107

Pijie V7.5: Wi 7e 178, 179-180

Fish bones, from excavations, 164-165
Fish gorges, 157-158, 173

Fish hooks, 157-158, 173

Foss, Dorothy B., 164

Gifford, Edward W., 17, 177, 178, 179
Gouges, shell, 154; stone, 137
Grace, George W., 174, 177, 180-181

Haas, Fritz, 162-163

Hawaii, 17, 177-179; University of, 180

Hornbostel, Gertude, 89-90

Hornbostel, Hans G., 8-9, 18, 20, 29, 35,
70, 73, 74, 84, 88-89, 96, 102, 137, 138,
142, 147, 151

Iron, Chamorro trade with Spanish for,
25, 168; artifacts, 165-167

Keesing, Felix M., 176
Kirchoff, Paul, 175

Knives, shell, 155; stone, 150
Kroeber, A; L.;51'75

Latte Phase, 20, 171-174

Libby,. W.. 2 .5:47,°65, 66, 177

Linguistics, significance for prehistory,
174, 177, 180-181

Magellan, 25, 167

Malaysia, 177, 180

Marche, A., 88

Marianas, climate, 24; land area, 22, 24;
vegetation, 24

Melanesia, 17-18, 175-181

Merrill, Elmer D., 181

Micronesia, archaeological status and
problems of, 17-20, 179; distribution
of pottery in, 18; prehistoric settlement
of, 176

Mortimer, George, 86

Mythology relating to House of Taga,
89-90

New Caledonia, 17, 177, 178, 179-180

Oceania, spatial framework for pre-
history of, 175-177; prehistoric chron-
ology of, 177-178; strategy of archae-
ological work in, 178-180

Ornaments, 147, 158-160, 173

Osborne, Douglas, 17

Pacific Science Board, 7

Palau Islands, 18, 20, 174, 179

Pendants, shell, 158-159, 160; stone, 147

Philippines, cultural affiliations with
Marianas, 20, 174-175

Physical anthropology, significance for
prehistory, 181

Pig, prehistoric, 25, 164, 181

Plants, cultivated, 25, 140, 174, 176, 181

Polynesia, 17-18, 174, 175-181

Population, estimate of prehistoric, 171—

Pottery, distribution in Micronesia, 18;
seriation of Marianas types, 123-127,
169-170; summary of Marianas types,
108

Radiocarbon dates, 65, 66, 84, 124, 168,
169, 170, 171, 177-178

186

a a ee a

INDEX 187

Sanchez y Zayas, 88

Sauer, Carl O., 181

Scrapers, shell, 155-157; stone, 148-150

Shells from excavations, 162-163

Social organization, ancient Chamorro
form of, 25-26, 98, 171-172

Soil profiles, 40, 50-51, 55, 60, 65-66,
78-82

Solheim, Wilhelm G., 115-116

Spanish conquest, 25, 35, 71, 107, 168

Spear points, 160, 164, 173

Stewart, T. Dale, 8, 84, 173

Stone quarrying, probable prehistoric
method used, 106

Stratigraphy, 57, 58, 98, 169-171

Thompson, Laura, 18, 25, 26, 29, 35, 74,
84, 89, 108, 114, 131, 138, 142, 145,
151-152, 157, 160, 169-170, 171

Trade materials, 165-167, 168
Tri-Institutional Pacific Program, 180-
181

U. S. Geological Survey, 8, 26, 66, 68-69
U. S. Navy, 7-8, 26, 35, 102

Warfare, ancient Chamorro, 25, 173

Wenner-Gren Foundation, 8

Willey, Gordon R., 175

Woods, Loren P., 164

Work areas, paved with coral pebbles, 36,
45-46

Yale University, 180
Yap, 17, 174, 179
Yaws, paleopathology of, 84, 181-182

a!
2
b
i
L
he
:
¢
:
in
“
Fi
fi
aa
:
rh
:
7 :

I tL ANN

12 084204681

Internet Archive Audio

Featured

Top

Images

Featured

Top

Software

Featured

Top

Books

Featured

Top

Video

Featured

Top

Mobile Apps

Browser Extensions

Archive-It Subscription

Save Page Now

Full text of "Marianas prehistory: archaeological survey and excavations on Saipan, Tinian and Rota"

See other formats