AMERICAN JOURNAL OF PHYSICAL ANTHROPOLOGY 131:1–14 (2006)
Paleodemography of a Medieval Population in Japan:
Analysis of Human Skeletal Remains from the
Yuigahama-minami Site
Tomohito Nagaoka,1* Kazuaki Hirata,1 Emi Yokota,2 and Shuji Matsu’ura2
1
Department of Anatomy, St. Marianna University School of Medicine, Miyamae Ward, Kawasaki,
Kanagawa 216-8511, Japan
2
Faculty of Human Life and Environmental Sciences, Ochanomizu University, Bunkyo Ward, Tokyo 112-8610, Japan
KEY WORDS
life expectancy; medieval population; skeletal remains; Japanese; secular trend
ABSTRACT
The purpose of this study is to obtain demographic data regarding the medieval population buried
at the Yuigahama-minami site in Kamakura, Japan, and to
detect a secular trend in the life expectancy of Japanese
population over the last several thousand years. The Yuigahama-minami skeletal sample consists of 260 individuals,
including 98 subadults (under 20 years old) and 162 adults.
A Yuigahama-minami abridged life-table analysis yielded a
life expectancy at birth (e0) of 24.0 years for both sexes, a life
expectancy at age 15 years (e15) of 15.8 years for males, and
an e15 of 18.0 years for females. The reliability of the estimated e0 was confirmed by analysis of the juvenility index.
Demographic profiles comparing the Yuigahama-minami
series with other skeletal series indicated that both the survivorship curve and life expectancy of the Yuigahama-min-
ami sample are similar to those of the Mesolithic-Neolithic
Jomon population, but are far lower than those of the early
modern Edo population. These comparisons strongly suggest that life expectancy changed little over the thousands
of years between the Mesolithic-Neolithic Jomon and medieval periods, but then improved remarkably during the few
hundred years between the medieval period and early modern Edo period. The short-lived tendency of the Yuigahamaminami sample does not contradict the archaeological hypothesis of unsanitary living conditions in medieval Kamakura. This is the first investigation to address the demographic features of a medieval population in Japan, and will
help refine our understanding of long-term trends in the demographic profiles of inhabitants of Japan. Am J Phys
Anthropol 131:1–14, 2006. V 2006 Wiley-Liss, Inc.
Paleodemographic studies provide important information regarding the life-history patterns of ancient populations, and are of great value for understanding population dynamics in historic and prehistoric times. The demographic reconstruction of ancient populations, which
has been conducted partially in the field of physical anthropology, has received attention since the early papers
of Angel (1947, 1954). Following these, studies of paleodemography became standard practice in physical anthropology, and continue to flourish (e.g., Kobayashi, 1967; Lovejoy, 1971; Lovejoy et al., 1977, Weiss, 1973; Bocquet-Appel
and Masset, 1982, 1985, 1996; Mensforth and Lovejoy,
1985; Nakahashi and Nagai, 1985, 1989; Gage and Dyke,
1986; Gage, 1988, 1989, 1990; Walker et al., 1998; Mensforth, 1990; Wood et al., 1992, 2002; Gage and Mode, 1993;
Konigsberg and Frankensberg, 1994; Alesan et al., 1999,
Drusini et al., 2001).
We are concerned in our research with the paleodemographic reconstruction of the ancient skeletal population in
Japan. We analyzed the demographic structure of medieval
skeletons from the Yuigahama-minami site (Kamakura,
Japan), and discuss their life-history patterns. This study
distinguishes itself from previous studies by focusing on
the paleodemography of the medieval Japanese.
Paleodemography has not yet been settled and has not
flourished in Japan. Indeed, we have only a few examples of paleodemographic analyses of Japanese skeletal
samples (Kobayashi, 1967; Nakahashi and Nagai, 1985,
1989). Kobayashi (1967) reconstructed demographic profiles of ancient populations in Japan, based on human
skeletal samples. Although Kobayashi (1967) investi-
gated the long-term life-span trends of inhabitants of
Japan from the prehistoric to modern periods, little is
known about the demographic features of populations other
than the Mesolithic-Neolithic Jomon (12,000–300 BC) and
early modern Edo populations (1600–1850 AD). Additional
demographic data were also reported for a medieval population (1300–1600 AD) (Nakahashi and Nagai, 1985), as
well as for an Aeneolithic population (300 BC–300 AD)
(Nakahashi and Nagai, 1989). However, these reports have
not enabled us to determine the long-term trends in life
span of the inhabitants of Japan.
The medieval demographic data collected by Nakahashi
and Nagai (1985) were insufficient for the elucidation of
a secular trend in the demographic features of inhabitants of Japan. Nakahashi and Nagai (1985) reported the
C 2006
V
WILEY-LISS, INC.
C
Grant sponsor: Ministry of Education, Culture, Sports, Science,
and Technology of Japan; Grant numbers: Grant-in-Aid for Scientific
Research on Priority Areas 15068102, Grant-in-Aid for Young Scientists
17770212.
*Correspondence to: Tomohito Nagaoka, Department of Anatomy,
St. Marianna University School of Medicine, 2-16-1 Sugao, Miyamae
Ward, Kawasaki, Kanagawa 216-8511, Japan.
E-mail: [email protected]
Received 4 June 2005; accepted 7 December 2005.
DOI 10.1002/ajpa.20402
Published online 27 January 2006 in Wiley InterScience
(www.interscience.wiley.com).
2
T. NAGAOKA ET AL.
life expectancy at birth of a medieval skeletal population
from the Yoshimohama site. This report was an important first description of the demographic aspects of a medieval Japanese population; however, the report lacked
some of the requirements for detailed demographic discussion. First, the report did not provide a profile of the
population structure of the Yoshimohama skeletal series
due to a lack of statistics. Second, Nakahashi and Nagai
(1985) did not present a life table of the Yoshimohama
population; thus, it was difficult to compare these data
with the data of Kobayashi (1967). Finally, the sample
size of the Yoshimohama population (57 subadult individuals and 50 adult individuals) was too small for demographic investigations.
Given the nature of previous studies, no satisfactory
discussion has taken place concerning the secular trend
in life span of inhabitants of Japan. The medieval population is a missing link in trying to trace long-term
trends in life span in Japan over the last several thousand years. To study this population, more medieval
skeletal samples need to be collected and investigated.
During the past several decades, many medieval skeletons were excavated from archaeological sites in the Yuigahama area of Kamakura City (Kanto District, Japan).
These excavations yielded approximately 5,000 individuals in varying states of preservation from the Zaimokuza
(556 individuals; Suzuki et al., 1956), Seiyokan (91 individuals; Morimoto et al., 1984), Yuigahama-minami (3,861
individuals; Hirata et al., 2002; Matsushita, 2002), and
Chusei Shudan Bochi (592 individuals; Hirata and
Nagaoka, no date) sites. Medieval Kamakura was an ancient capital where a military government, the Kamakura
Shogunate, was established. Archaeological data indicate
that the high population density in the city caused
increased competition and a prevalence of diseases related
to bad hygienic conditions (Kawano, 1989, 1995). Most
studies of skeletons excavated from these sites focused
on craniometry, odontometry, and paleopathology (Suzuki
et al., 1956; Morimoto et al., 1984; Hirata et al., 2002,
2004; Matsushita, 2002; Hirata and Nagaoka, no date;
Nagaoka and Hirata, 2006). Paleodemographic studies of
these skeletons will provide important information about
the mortality patterns and life conditions of this medieval
Japanese population.
The purpose of this study was to test the hypothesis
that health declined with increased aggregation and generally poor living conditions in the medieval period. We
do so by evaluation of changes in demographic structure
in this, in earlier, and in later settings in Japan. The
evaluation also enabled us to detect a secular trend in
the life expectancy of the inhabitants of Japan over the
past several thousand years.
MATERIALS
The materials utilized in this study consist of human
remains from the Yuigahama-minami site. The Yuigahama-minami site is located along the seashore of the
southern end of Kamakura City, and is adjacent to several other medieval sites: Zaimokuza, Seiyokan, and
Chusei Shudan Bochi (Fig. 1). The excavation of the Yuigahama-minami site was undertaken between 1995–
1997, and yielded 3,861 skeletons from the late 14th century layer (Hirata et al., 2002; Matsushita, 2002).
The skeletal samples used here were selected according to the following criteria. First, we selected samples
from individually buried graves. Individually buried
Fig. 1. Map indicating location of Yuigahama-minami site
and affiliated sites in Kamakura, Japan.
skeletons were found in articulated positions, and were
not mixed with the remains of other individuals. These
samples can be easily identified as single individuals.
Therefore, the age and sex of each sample can be determined synthetically from any available remains of the
individual involved.
Second, we selected only samples where the pelvis had
been preserved. Skeletal remains were found in varying
states of preservation; we attempted to control for preservation conditions using this second criterion. In addition, the pelvis contains much diagnostic information
concerning the age and sex of an individual. We studied
260 individual skeletons, housed in the Department of
Anatomy at St. Marianna University School of Medicine.
METHODS
Age determination
Skeletons were classified into 11 age categories: five
subadult groups (under 1 year, 1–4 years, 5–9 years, 10–
14 years, and 15–19 years) and six adult groups (20–24 years,
25–34 years, 35–44 years, 45–54 years, 55–64 years, and
65 years and over).
In this study, age at death was estimated on the basis of
the macroscopic observation of crania, teeth, ribs, pelvis,
and limb bones (Krogman, 1962; Brothwell, 1981; White
and Folkens, 2000). Age criteria for human skeletons are
usually constructed using a series of modern skeletons of
known age and sex. Thus, the assumption that age-related
changes in ancient skeletons are similar to those in modern
skeletons is required for age estimation.
3
PALEODEMOGRAPHY OF MEDIEVAL POPULATION IN JAPAN
TABLE 1. Composition by ages and sexes of
Yuigahama-minami sample
Subadult age determination
In the case of subadult skeletons, all aspects of dental
development, including the completeness of all crowns
and roots and the place of each tooth relative to the alveolar margin, were used for age determination (Fig. 71 in
Ubelaker, 1989). The degree of development and closure of
the occipital synchondrosis (Wakebe, 1990), and degree of
ossification and epiphyseal union of the pelvis and long
bones (Flecker, 1942; Webb and Suchey, 1985), were used
as secondary criteria.
Age group
0
1–4
5–9
10–14
15–19
Male
Female
Male
Female
Male
Female
Male
Female
Male
Female
Male
Female
Male
Female
20–24
25–34
Adult age determination
Estimation of age at time of death for each adult skeleton
was carried out using multiple criteria: stage of suture closure of the skull (Okada, 1961; Meindl and Lovejoy, 1985;
Suzuki, 1998), stage of tooth attrition (Miles, 1963; Lovejoy,
1985), chronological metamorphosis of the head, tubercle,
and costal face of the first rib (Charles, 1999), chronological
metamorphosis of the pubic symphysis (Hanihara, 1952),
chronological metamorphosis of the auricular surface of the
ilium (Lovejoy et al., 1985), and the presence of osteophytes
and arthritic lesions in the vertebrae (Erickson, 1978;
Nathan, 1962; Stewart, 1958).
35–44
45–54
55–64
65–
Total
Total
Total
Total
Total
Sex
subadults
adults
males
females
N
8
37
17
15
13
8
20
18
36
29
16
20
9
9
1
4
0
0
98
162
95
88
260
Sex determination
No attempt was made to determine the sex of subadult
individuals aged 14 years and under. Sex determination
of individuals aged 15 years and greater was carried out
based on macroscopic assessment of pelvic features.
Dimorphic criteria of the pelvis included the greater sciatic notch (Genovés, 1959; Ferembach et al., 1980; Bruzek, 2002), preauricular sulcus (Houghton, 1974; Ferembach et al., 1980; Bruzek, 2002), ventral arc (Phenice,
1969), subpubic concavity (Phenice, 1969), and medial
aspect of the ishiopubic ramus (Phenice, 1969).
Demographic methodology
According to Alesan et al. (1999), the following assumptions should be made in undertaking a demographic study
of a population. First, only one population, community, or
group utilized the cemetery from which the remains were
excavated, and this group did not concurrently use other
cemeteries. Second, all individuals from that community
were buried in the necropolis. Third, the archaeological
excavation was complete, and anthropological recovery was
meticulous and not differential. Finally, entrances by birth
balanced exits by death, and there were no differences
between these events in terms of age or sex. We assumed
that our skeletal population was representative of the living
population who died in this community. Without these
assumptions, paleodemographic reconstructions are not possible (Alesan et al., 1999). In this study, we also assumed
that skeletal preservation conditions did not differ based on
age or sex.
A life table allows us to determine a statistical description of mortality parameters. The use of a life table
implies two assumptions (Alesan et al., 1999): 1) that the
obtained age-at-death structure is real, and 2) that the
population is stationary. In this case, we believe that
these assumptions are appropriate, because ancient populations are thought to have had growth rates of virtually zero, and to have been more stable as far as mortality is concerned. This study compares our demographic
data from the Yuigahama-minami site with the model
life tables of Weiss (1973), so that we may determine an
age pattern of mortality that we would not be able to
describe from the original data. Although model life
tables are powerful tools for paleodemographic analyses,
they are subject to potential biases resulting from the
use of inappropriate model populations (Gage, 1988).
An alternative to model life tables is mathematical
hazard models (Gage, 1988, 1989, 1990; Gage and Dyke,
1986; Gage and Mode, 1993; Wood et al., 1992, 2002).
The use of hazard models of mortality for paleodemographic analysis allows us to determine reliable estimates for mortality at all ages, even at those ages for
which precise age estimates are impossible (Milner
et al., 2000). One of the most parsimonious parametric
models of mortality across the entire life span is the
Siler competing hazards model, developed for animal
populations by Siler (1979, 1983) and extensively tested
by Gage and Dyke (1986) and Gage and Mode (1993).
This model fits as well as, or better than, most of the
other models for human mortality data (Gage and Dyke,
1986; Gage and Mode, 1993). As shown (Gage, 1988,
1989, 1990; Gage and Dyke, 1986; Gage and Mode,
1993), the mortality function of the model is
mx ¼ a1 expðb1 xÞ þ a2 þ a3 expðb3 xÞ
where mx is the hazard function for paleodemographic
reconstruction, or the force of mortality resulting from
all three competing hazards at exact age x, and where
each of the three additive terms on the right-hand side
represents a different component of mortality. The first
additive component of the equation is a negative Gompertz equation, and represents the rapid decline in mortality during the first few years of life. This is referred
to as the immature component of mortality. The parameter a1 is the force of mortality resulting from immaturity
at the moment of birth, and b1 is the rate at which
immature mortality decreases with respect to age. The
second component, a2, represents a force of mortality
that is constant with respect to age. This component was
4
T. NAGAOKA ET AL.
TABLE 2. Yuigahama-minami sex-specific survivorship1
Total
Male
Female
x (age)
N
Percent
lx
N
Percent
lx
N
Percent
lx
15–19
20–24
25–34
35–44
45–54
55–64
Total
21
38
65
36
18
5
183
11.5
20.8
35.5
19.7
9.8
2.7
1,000.0
885.2
677.6
322.4
125.7
27.3
13
20
36
16
9
1
95
13.7
21.1
37.9
16.8
9.5
1.1
1,000.0
863.2
652.6
273.7
105.3
10.5
8
18
29
20
9
4
88
9.1
20.5
33.0
22.7
10.2
4.5
1,000.0
909.1
704.5
375.0
147.7
45.5
1
lx, number of survivors to age x, based on radix of 1,000.
TABLE 3. Yuigahama-minami abridged life table combined for both sexes1
x (age)
Dx
lx0
lx
dx
qx
Lx
Tx
ex
0
1–4
5–9
10–14
15–19
20–24
25–34
35–44
45–54
55–64
Total
8
37
17
15
21
38
65
36
18
5
260
260
252
215
198
183
162
124
59
23
5
1,000.0
969.2
826.9
761.5
703.8
623.1
476.9
226.9
88.5
19.2
30.8
142.3
65.4
57.7
80.8
146.2
250.0
138.5
69.2
19.2
0.031
0.147
0.079
0.076
0.115
0.235
0.524
0.610
0.783
1.000
984.6
3,592.3
3,971.2
3,663.5
3,317.3
2,750.0
3,519.2
1,576.9
538.5
96.2
24,009.6
23,025.0
19,432.7
15,461.5
11,798.1
8,480.8
5,730.8
2,211.5
634.6
96.2
24.0
23.8
23.5
20.3
16.8
13.6
12.0
9.7
7.2
5.0
1
x, ages in years; Dx, absolute number of individuals dead at age x; l0 x, absolute number of survivors to age x; lx, number of survivors to
age x, based on radix of 1,000; dx, number of dead persons to age x, based on radix of 1,000; qx, probability of dying in succeeding age
class for those who reach age x; Mx, age-specific death rate at age x; Lx, number of person years lived in age class x; Tx, total number of
person years to be lived by population at age x before all are dead; ex, number of years a person can expect to live beyond age x.
first proposed by Makeham (1860) as an addition to the
Gompertz mortality model. A baseline hazard parameter
(a2) was excluded from this study, as this parameter is
rarely estimated from paleodemographic data (Herrmann
and Konigsberg, 2002). The third additive component, the
Gompertz mortality model, which has been used extensively
to smooth and improve mortality estimates over adult age
groups, is referred to here as the senescent component of
mortality. The parameter a3 is the force of mortality resulting from senescence at the moment of birth, and b3 is the
rate at which this force of mortality increases with age. The
Siler competing hazards model is ideally suited to smoothing and improving demographic data, including age-specific
survivorship rates from paleodemographic populations,
because it smoothes the data without imposing a particular
age pattern of mortality (Gage, 1988). The Siler parameters
are estimated in mle version 2.1 (Holman, 2003).
To estimate certain demographic parameters (e0, life expectancy at birth; 1q0, probability of dying between birth
and age 1 year; and 5q0, probability of dying between birth
and age 5 years), the juvenility index (Bocquet-Appel and
Masset, 1996) was calculated as the ratio between the number of deaths between ages 5–14 years and the number of
deaths after age 20 (D5–14/D20–w). This index allows us to
estimate demographic parameters without the bias due to
infant underrepresentation in osteological collections, and
also to control for systematic bias in the calculation of adult
age distribution.
RESULTS
Age and sex composition
of Yuigahama-minami sample
The composition by age and sex of the studied skeletal
series is summarized in Table 1. The series comprises
260 individuals, including 98 subadults (under 20 years
old) and 162 adults. More specifically, 8 individuals were
younger than 1 year, 37 individuals were aged 1–4 years,
17 individuals were aged 5–9 years, 15 individuals were
aged 10–14 years, 21 individuals were aged 15–19 years,
38 individuals were aged 20–24 years, 65 individuals were
aged 25–34 years, 36 individuals were aged 35–44 years,
18 individuals were aged 45–54 years, 5 individuals were
aged 55–64 years, and 0 individuals were classified as over
65. The number of individuals trended lower from the 25–
34-year group onward. The mean age of the population was
estimated at 24.0 years.
Sex-specific age distributions in the age range of 15 years
and over for the Yuigahama-minami population are listed
in Table 2. The overall sex ratio was 106 males to 100 females.
The sex ratios computed by age group were 163:100 in the
15–19-year group, 111:100 in the 20–24-year group, 124:100
in the 25–34-year group, 80:100 in the 35–44-year group,
100:100 in the 45–54-year group, and 25:100 in the 55–
64-year group. There was a higher proportion of males in
the 15–19-year to 25–34-year age groups, and a higher proportion of females in the 35–44-year to 55–64-year age
groups. The mean age of the population of individuals older
than 15 years was 30.6 years for males, 33.0 years for
females, and 31.8 years for all individuals.
Demographic parameters of Yuigahama-minami
population, based on life tables
Table 3 shows the life table computed for the entire
population. Figures 2–5 show curves representing the number of surviving individuals (lx), number of dead individuals
(dx), mortality rate (qx), and life expectancy (ex), respectively, based on the data shown in Table 3.
PALEODEMOGRAPHY OF MEDIEVAL POPULATION IN JAPAN
Fig. 2. Number of survivors for Yuigahama-minami site
(combined sexes).
Fig. 3. Number of dead for Yuigahama-minami site (combined sexes).
Examination of the life table for the entire population
(Table 3) demonstrates the following points. First, the
number of survivors decreased gradually: the number of
survivors aged 25–34 was half the number at birth, and
the number of survivors aged 35–44 was a quarter of the
number at birth. Second, the number of dead peaked
slightly at ages 1–4 and, remarkably, peaked again at
ages 25–34. Third, the mortality rate peaked slightly at
ages 1–4, subsequently decreased, and then increased
gradually after ages 15–19. Finally, life-table analysis of
the Yuigahama-minami population showed that life expectancy decreased with age. Specifically, the estimated life
expectancy was 24.0 years at birth, 16.8 years at ages 15–
19, 9.7 years at ages 35–44, and 5.0 years at ages 55–64.
Table 4 shows the sex-specific life table for the age
range of 15 years and over. Figures 6–9 show curves representing the number of surviving individuals (lx), number of dead individuals (dx), mortality rate (qx), and life
expectancy (ex), respectively.
Examination of the sex-specific life table demonstrates
the following points. First, the number of survivors with
age decreased more rapidly among males than females.
Second, the number of deaths peaked at the 25–34-year
age group for both sexes, but the peak was higher for
males than females. The number of dead was larger for
5
Fig. 4. Probability of dying for Yuigahama-minami site
(combined sexes).
Fig. 5. Life expectancy for Yuigahama-minami site (combined sexes).
males than females before age 34, but was smaller for
males than females after age 35. Third, while the mortality rate increased gradually with age for both sexes,
the mortality rate for males was higher than that for
females beginning at age 15 and continuing throughout
all of the older age groups. Finally, life expectancy decreased with age, but females had a longer life expectancy than males by several years at all age groups.
Comparison of Yuigahama-minami population
and models of Weiss (1973)
Comparison of the demographic data from the Yuigahama-minami population with the model life tables of
Weiss (1973) provides an age pattern of mortality that
could not be ascertained by analysis of the original data
alone. The models were chosen based on two entrance
parameters, e15 and l15. The models of e15 ¼ 15 and l15 ¼
70, on the one hand, and e15 ¼ 20 and l15 ¼ 70, on the
other, were chosen based on the entrance parameter values obtained from Table 3. In the comparison of life-table
parameters, lx, from the Yuigahama-minami population
with the model life table, two main biases were identified: a deficit of individuals under age 1 year and a defi-
6
T. NAGAOKA ET AL.
TABLE 4. Yuigahama-minami sex-specific abridged life table1
x (age)
Male
15–19
20–24
25–34
35–44
45–54
55–64
Total
Female
15–19
20–24
25–34
35–44
45–54
55–64
Total
1
Dx
lx0
lx
dx
qx
Lx
Tx
ex
13
20
36
16
9
1
95
95
82
62
26
10
1
1,000.0
863.2
652.6
273.7
105.3
10.5
136.8
210.5
378.9
168.4
94.7
10.5
0.137
0.244
0.581
0.615
0.900
1.000
4,657.9
3,789.5
4,631.6
1,894.7
578.9
52.6
15,605.3
10,947.4
7,157.9
2,526.3
631.6
52.6
15.6
12.7
11.0
9.2
6.0
5.0
8
18
29
20
9
4
88
88
80
62
33
13
4
1,000.0
909.1
704.5
375.0
147.7
45.5
90.9
204.5
329.5
227.3
102.3
45.5
0.091
0.225
0.468
0.606
0.692
1.000
4,772.7
4,034.1
5,397.7
2,613.6
965.9
227.3
18,011.4
13,238.6
9,204.5
3,806.8
1,193.2
227.3
18.0
14.6
13.1
10.2
8.1
5.0
See Table 3 for explanation of symbols.
Fig. 6. Sex-specific curves of number of survivors for Yuigahama-minami site.
Fig. 7. Sex-specific curves of number of dead for Yuigahama-minami site.
cit of elderly individuals (Fig. 10). The first bias observed
was the underrepresentation of infants. The data from
the life table suggest that only 3.1% of individuals would
have died during the first year of life, while the models
by Weiss (1973) of both MT (Model Table) 15–70 and
Fig. 8. Sex-specific curves of probability of dying for Yuigahama-minami site.
Fig. 9. Sex-specific curves of life expectancy for Yuigahamaminami site.
MT 20–70 yielded an expected value of 13.3%. This indicates a deficit of 35 individuals under age 1 year. The
second bias identified was a deficit of elderly remains
relative to the prediction of the models. The loss of individuals over age 55 was 6 individuals in the case of MT
15–70, and 26 individuals in the case of MT 20–70.
PALEODEMOGRAPHY OF MEDIEVAL POPULATION IN JAPAN
7
Fig. 10. Comparison of Yuigahama-minami and model of
Weiss (1973) (MT 15–70 and MT 20–70) (number of survivors).
Fig. 12. Fitted and original age-specific survivorship curves
for Yuigahama-minami site (number of survivors).
Demographic parameters of Yuigahama-minami
population, based on juvenility index
Fig. 11. Siler hazard function for paleodemographic reconstruction of Yuigahama-minami site.
Siler competing hazards model for
Yuigahama-minami population
The Siler competing hazards model contained all parameters except the a2 parameter. The following parameter
estimates were calculated: a1 ¼ 0.054, b1 ¼ 0.366, a3 ¼
0.011, and b3 ¼ 0.058. Figures 11 and 12 show the Siler
hazard function and the fitted and original age-specific
survivorship curves, respectively. The Siler hazard function for paleodemographic reconstruction of the Yuigahama-minami series provided a plausible age pattern of
mortality, due to the observation that the instantaneous
death rate declined swiftly and continuously during the
first few years of life, reached a minimum around ages
10–15, and then increased at an increasing rate.
In Figure 12, similar trends are seen in both the fitted
and original survivorship curves. However, the graduated
curve declined a bit more rapidly than the original curve.
The number of survivors at age 24 was half the number at
birth, and the number of survivors at age 36 was a quarter
of the number at birth (Fig. 12). The fitted survivorship
curve seemed to provide a more plausible age pattern of mortality than the original curve, as the fitted curve conformed
to the general characteristics of human mortality patterns,
due to a slightly higher probability of dying at birth.
Bocquet-Appel and Masset (1996) derived a new set of
paleodemographic estimators using the juvenility index
at death (D5–14/D20–w). These estimators yield the life expectancy at birth (e0) and the probability of death at 1 and
5 years (1q0 and 5q0).
Table 5 shows the paleodemographic estimators calculated for the Yuigahama-minami sample. The data from
the life table suggest that only 3.1% of individuals would
have died within the first year of life, and 17.6% of individuals would have died in their first 5 years of life. However,
when estimated using the juvenility index, the values obtained for 1q0 and 5q0 were 27.6 6 1.5% and 44.1 6 1.5%,
respectively. Therefore, the life table underestimates the
rate of infant mortality.
The life-table analysis yielded a life expectancy at birth
of 24.0 years. The life expectancy at birth estimated from
the juvenility index was 24.8 6 1.5 years (Table 5). This
estimation is in good agreement with the value calculated
from the life table. No significant discrepancy was found
between the two approaches for estimating life expectancy.
Therefore, the juvenility-index analysis confirmed the reliability of the estimated life expectancy from the life table.
DISCUSSION
Census errors in Yuigahama-minami sample
Age-at-death structure. Skeletal series unearthed frequently display age-at-death patterns that differ from
those of typical living populations. In our comparison of
the life-table parameters of the Yuigahama-minami site
with the models of Weiss (1973), the Siler hazard model,
and the parameters estimated using the juvenility index,
we observed that children younger than 5 years are underrepresented in our population. We do not actually know the
infant mortality rate of an ancient population, but according to observed patterns in preindustrial populations (e.g.,
Howell, 1979), we should expect high infant mortality. A
higher probability of death during the early years is expected for endogenous and exogenous reasons, and due to a
lack of medical and hygienic practices in such a population.
8
T. NAGAOKA ET AL.
TABLE 5. Demographic parameters computed from life table and juvenility index
Series
Chronology
Juvenility
index
D5–14
D20–w
e01
e02
1
1q0
2
1q0
1
5q0
2
5q0
Yuigahama-minami
14th century AD
0.198
32
162
24.010
24.767
0.031
0.276
0.173
0.441
1
2
Computed from life table.
Computed from juvenility index.
TABLE 6. Basic statistics of MCH for Yuigahama-minami samples
Male
Female
Measurement item
N
Mean (mm)
SD
N
Mean (mm)
SD
t-value
Probability
MCH
75
61.7
4.13
65
54.4
3.87
7.87
P < 0.001
Infant underrepresentation is often cited as a major source
of census errors in age distributions that are reconstructed
from both extant and extinct human groups (Angel, 1969;
Weiss, 1973; Bocquet-Appel and Masset, 1982, 1985; Mensforth, 1990; Alesan et al., 1999). As Mensforth (1990)
pointed out, selective cultural biases and mortuary practices at time of death, differential postmortem preservation,
selective recovery, and variation in the degree to which age
and sex can accurately be inferred from fragmentary skeletal remains all increase the vulnerability of skeletal series
to sampling errors. Cultural behaviors that promote this
bias include preferential mortuary practices, in which the
very young are buried separately from adults, and the practice of infanticide. Furthermore, skeletal populations are at
greater risk of infant underrepresentation due to the fact
that infant skeletal remains are smaller and less well-mineralized, and thus are more susceptible to postmortem
physical and chemical deterioration (Gordon and Buikstra,
1981; Mensforth, 1990; Mays, 1998).
Another possible bias identified in the life table of the
Yuigahama-minami population is a deficit of elderly remains relative to predictions from model life tables. The
life-table approach is open to criticism on several grounds,
including poorly defined age-at-death distribution and age
estimation errors (Wood et al., 2002). We cannot exclude
the possibility that methodological factors related to difficulties in determining age could be an underlying cause of
this bias (Bocquet-Appel and Masset, 1982, 1985, 1996;
Walker et al., 1988; Mensforth, 1990; Konigsberg and
Frankensberg, 1994). However, despite these criticisms and
in contrast with what happens with infants, the deficit of elderly individuals could reflect the actual state of the population. Paleopathological data support the hypothesis that
this population existed under difficult living conditions.
Cases of cribra orbitalia, cribra cranii, tuberculosis, leprosy,
osteoarthritis, fracture, and injury by weapon (Hirata et al.,
2002) are some examples of the supporting paleopathological evidence on this issue. The deficit of elderly individuals
may reflect a population with a low probability of reaching
old age. According to the overall small number of individuals in old age groups in archaeological samples, the effect of
underestimation of elderly remains might be less than
expected, as suggested by Nagar and Hershkovitz (2004).
If this deficit of elderly individuals reflects the actual
situation, comparing our calculated life-table parameters
with the parameters of other skeletal series would be
beneficial in attempting to detect a secular trend in the
demographic features of inhabitants of Japan. The use of
both original life-table estimations and the Siler competing hazards model will facilitate the empirical comparison of age patterns of survivorship among several popu-
Fig. 13. Percent distribution of male and female MCH measurements, with respect to cutpoint determined by discriminant
function based on MCH alone. Individuals to right of cutpoint
would be classified as males, and individuals at left would be
classified as females.
lations, even for those ages at which precise age estimates are impossible (Milner et al., 2000).
Sex estimation. This study showed sexual differences
in life expectancy in the Yuigahama-minami sample.
However, one might wonder whether the age-related sex
estimation bias played a part in the analysis. We should
examine the reliability of sex determination, particularly
in the case of elderly individuals.
Discriminant function analysis helps us to assess errors
in sex determination. Some studies demonstrated the importance of the diaphyseal circumferences of long bones in
the diagnosis of sex (Black, 1978; Is˛can et al., 1994; Safont
et al., 2000). The minimum circumference of the humerus
(MCH), one of the most effective indicators of sex (Safont
et al., 2000), was utilized to calculate a discriminant function of sex. The sex of the Yuigahama-minami sample used
in the discriminant analysis had already been determined
by macroscopic observation of pelvic bones. None of the
humeri used in this analysis showed any pathology or
anomaly that distorted the dimensions of the diaphyses.
The definition of MCH was described in the classical anthropology literature (no. 7, in Martin and Knussmann,
1988, p. 199). The discriminant function was computed
with SPSS for Windows, version 13.0J (SPSS, Inc., 2005).
Table 6 contains descriptive statistics, including means
and standards deviations of MCH. The MCH of males
was clearly larger than that of females (t ¼ 7.87, P <
0.001) (Fig. 13). Table 7 shows the discriminant function
9
PALEODEMOGRAPHY OF MEDIEVAL POPULATION IN JAPAN
TABLE 7. Direct discriminant function calculated with MCH for Yuigahama-minami samples
N
Discriminant
function
Wilks’
lambda
0.249 * MCH-14.530
0.543
Eigenvalue
Canonical
correlation
Cutpoint
(mm)
Male
Female
0.843
0.676
58.3
63
75
Correct classification
Total
Male
(%)
Female
(%)
Total
(%)
138
82.2
89.2
85.5
TABLE 8. Number of misclassified individuals
by discriminant function
N
Misclassification
x (age)
Male
Female
Male
Female
15–19
20–24
25–34
35–44
45–54
55–64
Total
7
18
26
14
7
1
73
3
14
21
16
8
3
65
2
4
4
2
0
1
13
1
1
2
3
0
0
7
and a number of statistics. The discriminant function shows
a Wilk’s k of 0.543, an eigenvalue of 0.843, a canonical correlation of 0.676, and percentages of correct classification of
82.2% for males, 89.2% for females, and 85.5% for both
sexes. The number of misclassified individuals was 20.
Except for one male in the 55–64-year age group, these individuals were all younger than age 45 (Table 8). The frequency of misclassification was not significantly different
between males and females (v2 ¼ 2.42, P > 0.05). The individuals identified as misclassified by the discriminant function were classified into the opposite sex group by the macroscopic observation of pelvic bones. The result of this analysis does not contradict the longer life expectancy of females,
as the elderly females in the 45–54-year and 55–64-year age
groups are likely to be correctly classified.
Comparison of demographic profiles of
Yuigahama-minami and other skeletal series
As presented above, this study determined the demographic profile of the Yuigahama-minami sample. Below,
we compared our data with data obtained from other
skeletal series (Kobayashi, 1967) (Table 9). Kobayashi
(1967) computed sex-specific life tables based solely on
data from skeletons of individuals estimated at age 15
and older. The exclusion of skeletons of individuals estimated at younger than age 15 circumvents the unavoidable problem of underrepresentation of infants in paleodemographic populations. For the purpose of comparison,
we should make the assumption that errors in age and
sex estimation are negligible between different observers. Indeed, the methodology of age and sex estimation
used in this work is in part common to that of Kobayashi
(1967), who used the traditional age and sex estimation
indicators described in classic anthropology textbooks
(e.g., Krogmann, 1962).
Figure 14 shows the curves representing the number
of survivors (lx) in the age range of 15 and over for Yuigahama-minami medieval males (this study), MesolithicNeolithic Jomon males (Kobayashi, 1967), and early
modern Edo males (Kobayashi, 1967). Comparison of the
survival curves of these groups shows that the number
of survivors aged 35–44 is 27.4% of the number of survivors aged 15–19 in the medieval population, 28.6% in
the Jomon population, and 65.8% in the Edo population.
The number of survivors aged 55–64 is 1.5% in the medieval population, 1.2% in the Jomon population, and 32.5%
Fig. 14. Comparison of Yuigahama-minami males with Mesolithic-Neolihtic Jomon males and early modern Edo males (number
of survivors).
in the Edo population. The medieval and Jomon populations have far lower numbers of male survivors relative to
the Edo population at all age stages after age 15. The survivorship curve of Yuigahama-minami medieval males is virtually identical to that of Jomon males, but very different
from that of Edo males. Indeed, this tendency was detected
not only in males but also in females (Fig. 15).
The sex-combined survivorship curve fitted by the
Siler competing hazards model (Fig. 12), which facilitated the empirical comparison of age patterns among
several populations, also strongly confirms this trend. At
all age stages after age 15, the fitted Yuigahama-minami
data show virtually identical numbers of survivors when
compared to the Jomon data; in contrast, the number of
survivors is much higher in the Edo population (Fig. 16).
Figures 17 and 18 show the respective life-expectancy
curves for males and females in three populations. Our
analysis estimated a life expectancy of 15.6 years for
males and 18.0 years for females at ages 15–19 in the
Yuigahama-minami population. According to Kobayashi
(1967), the estimated life expectancy at ages 15–19 in the
Jomon samples was 16.2 years for males and 16.3 years for
females, while that in the Edo samples was 30.3 years for
males and 25.6 years for females. The life expectancy in the
Yuigahama-minami series for both sexes is almost similar
to that in the Jomon samples, but is half that in the Edo
samples, both in males and females. This difference persists
from age 15 throughout all greater age groups, as shown in
Figures 17 and 18.
Comparison of these demographic profiles demonstrated
similarities between the Jomon and Yuigahama-minami
medieval populations, and dissimilarities between the
Yuigahama-minami and Edo populations. We note that life
expectancy did not change significantly during the thousands of years between the Mesolithic-Neolithic Jomon and
medieval periods, but then doubled during the few hundred
years between the medieval and early modern Edo periods.
10
T. NAGAOKA ET AL.
Fig. 15. Comparison of Yuigahama-minami females with
Mesolithic-Neolihtic Jomon females and early modern Edo
females (number of survivors).
Fig. 16. Comparison of Yuigahama-minami curve (sex-combined) fitted by Siler hazard model with Mesolithic-Neolihtic
Jomon and early modern Edo males and females (number of
survivors).
Life history reconstructed from
Yuigahama-minami sample
A comparison of life-history data of the Yuigahamaminami sample with those of other skeletal series
showed two clear contrasts. First, the Yuigahama-minami sample differed from other skeletal samples in that
higher mortality was found in males than in females. Second, the Yuigahama-minami sample demonstrated a relatively short life expectancy when compared to the early
modern Edo sample.
Sexual difference in life expectancy in the Yuigahamaminami sample. In the Yuigahama-minami sample,
males had a life expectancy that was several years shorter
than for females. Interestingly, shorter life expectancy for
males was not observed in either the Mesolithic-Neolithic
Jomon sample or early modern Edo sample (Kobayashi,
1967). The data for the Jomon and Edo periods compiled
Fig. 17. Comparison of Yuigahama-minami males with Mesolithic-Neolihtic Jomon males and early modern Edo males (life
expectancy).
Fig. 18. Comparison of Yuigahama-minami females with
Mesolithic-Neolihtic Jomon females and early modern Edo females
(life expectancy).
by Kobayashi (1967) showed no sexual differences in life
expectancy. In contrast, according to a study of historical
documents (shumon-aratamecho) by Kito (2000), females
had a shorter life expectancy than males in the early
modern Edo period. The higher mortality rate for females
was centered between ages 20–40, an age span that corresponds to the reproductive period, gestation, childbirth,
and nursing (Kito, 2000). The sexual difference in mortality patterns in the Yuigahama-minami sample raises
the question, ‘‘What accelerated the death of males in the
medieval period?’’
A pioneering study on the medieval population of Japan by Suzuki et al. (1956) provides a possible explanation. Suzuki et al. (1956) excavated 556 skeletons from
multiple burials at the Zaimokuza medieval site. Most of
the dead were male adults, and they frequently had cut
marks that may have been caused by sharp weapons,
such as Japanese samurai swords. Based on these observations, Suzuki et al. (1956) assumed that the excavated
11
PALEODEMOGRAPHY OF MEDIEVAL POPULATION IN JAPAN
TABLE 9. Secular trend in life expectancy calculated based on ancient skeletons in Japan
e151
Series
Chronology
Male
Female
Male þ
Female
Misolithic-Neolithic
Jomon
Aeneolithic Yayoi
12,000–300 BC
16.20
16.30
16.20
Yuigahama-minami
Medieval
Yoshimohama
Medieval
Early Modern Edo
1
2
300 BC–300 AD
1300–1400 AD
e02 male þ
female
22.80
15.60
18.00
16.80
1300–1600 AD
1600–1850 AD
e01 male þ
female
24.01
20.60
30.30
25.60
27.82
24.77
Reference
Kobayashi
(1967)
Nakahashi and
Nagai (1989)
Present
study
Nakahashi and
Nagai (1985)
Kobayashi
(1967)
Computed from life table.
Computed from juvenility index.
skeletons were victims of a war. This assumption may
also explain the higher mortality of Yuigahama-minami
males, as the excavated skeletons from the Yuigahamaminami site also occasionally showed evidence of injuries
in the cranial and postcranial bones (Hirata et al., 2002,
2004).
However, the assumption by Suzuki et al. (1956) is difficult to apply directly to the Yuigahama-minami skeletons.
Many more individually buried skeletons were found at
the Yuigahama-minami site than at the Zaimokuza site.
Among these skeletons, the male-to-female sexual ratio
was approximately equal, and injured skeletons were found
much less frequently. The possibility that the males were
involved directly or indirectly in a war cannot be entirely
ruled out, but with the limited data in hand, the causes of
death remain unresolved. We believe that it is difficult to
attribute the higher mortality rate of males specifically to
war. Further analysis is necessary regarding the causes of
death of the Yuigahama-minami population.
Life expectancy in Yuigahama-minami sample and
their lives. The life expectancies of the Yuigahamaminami and Jomon samples were almost similar, and
were about half of the life expectancy of the Edo population. The short-lived tendency in the Yuigahama-minami
sample is likely to be connected to the living conditions
in medieval Kamakura, as described below.
The Yuigahama-minami individuals were inhabitants of
medieval Kamakura, an ancient capital where a military
government, the Kamakura Shogunate, was established.
Kamakura was also a center of politics, economics, and culture during the medieval period of Japan. According to
Kawano (1989), medieval Kamakura had a population of
between 70,000–100,000, and the population density in the
medieval period was higher than in the present day. The
overpopulated conditions may have impacted negatively on
the lives of the inhabitants. Analysis of archaeological
remains showed that the bodies of horses and humans were
often abandoned in side ditches or on streets in Kamakura
(Kawano, 1995). Scenes of dogs devouring abandoned
human remains and people excreting in the streets are portrayed on the medieval picture scroll Gakizoshi. In fact, we
found markings consistent with carnivore-gnawing on
some of the Yuigahama-minami sample (Hirata et al.,
2002). We can assume that sanitary facilities were not wellequipped, and that waste and cadavers were left on streets.
Additionally, cases of cribra orbitalia, cribra cranii, tuberculosis, leprosy, osteoarthritis, fracture, and injury by weapon
were reported in the Yuigahama-minami sample by Hirata
et al. (2002). The observation of injured skeletons implies
the possibility that some of the Yuigahama-minami individuals were drawn into direct or indirect involvement in violence or war. However, diseases that leave no diagnostic evidence in bones, including typhus, cholera, and dysentery,
cannot be identified by paleopathological investigation.
Comparison of the living conditions described above
with living conditions in the early modern Edo period
produces a marked contrast. The early modern Edo sample studied by Kobayashi (1967) was of urban inhabitants of the city of Edo (the old name of Tokyo) in the
latter half of early modern times. The period between
medieval and early modern times corresponds to dramatic
changes in urban environments in Japan. Edo became a
major city around the beginning of the 17th century under
the Tokugawa Shogunate. As previously reported (Hanley,
1987), the level of Japan’s metropolitan sanitation from the
mid-17th century to mid-19th century surpassed that of the
West, both in terms of water supply and waste disposal.
Clean drinking water was constantly supplied from aqueducts, and excrement was often recycled into manure for
agricultural use. Customs concerning hygiene, and food
and drink, combined with a lack of domestic animals, suggest that living conditions in Edo were more sanitary than
in Western cities. Furthermore, there is evidence that
standards in medicine began to improve during the early
modern Edo period. Vaccination was introduced in Japan
during the early modern Edo period (Sakai, 1982). Japanese
medicine was modernized following the epoch-making publication by Genpaku Sugita and other doctors of Kaitai
Shinsho, the first translation of a Western anatomy book
(Sakai, 1982).
Comparison of the lives and living conditions of the medieval Kamakura population and the early modern Edo population suggests that the time between the medieval and
early modern periods corresponded not only to a span of remarkable extension of life expectancy, but also to a span of
improvement of living conditions, including sanitary facilities and medical care. With the limited data in hand, the
main causes of death remain obscure, but at least the unsanitary living conditions during the medieval period certainly do not contradict the short-lived tendency of the Yuigahama-minami samples.
Japanese life expectancy: past and present
In this study, we investigated past Japanese life expectancy on the basis of ancient skeletons. Our data can be
used to refine ideas about a secular trend in life expectancy
12
T. NAGAOKA ET AL.
of the inhabitants of Japan. These paleoanthropological
data cannot be easily compared with data obtained by
national census. However, in spite of differences in demographic methodology, a comparison is useful for the purpose
of examining the secular trend in life expectancy in Japan.
A Japanese national census was first undertaken in
the late 19th century. According to the 1891–1898 national
survey, the life expectancy at birth was 35.3 years for males
and 35.8 years for females (Mizushima, 1961). Until the
end of World War II, people in Japan lived to around 50.
High rates of infant mortality contributed to a lower life expectancy (Fukawa, 2002). However, just a half century after World War II, the Japanese have become one of the longest-lived people in the world. The life expectancy at birth
in 2003 was 78.4 years for males and 85.3 years for females
(Ministry of Health, Labor and Welfare, 2004). The fact
that Japan has not been involved in any wars during this
period is one reason for the increased life expectancy. Other
factors include higher standards of nutrition and hygiene,
advances in medical technology, and easy access to medical
services, which have saved the lives of many newborns and
people of older ages (Fukawa, 2002). These census data
indicate that Japanese life expectancy has increased since
the late 19th century.
When does the beginning of the increasing trend in
Japanese life expectancy date to? Interestingly, consideration of previous studies and the present analysis suggests that the beginning of improvement of life expectancy dated to before the 19th century. Our data show
that the life expectancy doubled for all age groups after
age 15 during the hundreds of years between the medieval and early modern Edo periods. In the demographic
profiles of Kobayashi (1967), who compared osteological
data in the Edo period with national census data in the
late 19th century, life expectancy was improved in all age
groups after age 15 from the early modern Edo period to
the late 19th century. Additionally, in the study of historical
documents (shumon-aratamecho), Kito (2000) also indicated that life expectancy in all age groups after age 2 was
improved by several years within the early modern Edo period between the 17th and mid-19th centuries.
In conclusion, Japanese life expectancy has improved
from the medieval to the early modern and modern periods, and even within the early modern period. Thus, we
infer that Japanese life expectancy began to increase following the medieval period, and that this increasing
trend in Japanese life expectancy has continued to the
present.
CONCLUSIONS
The analysis of demographic data for the Yuigahamaminami population has refined our understanding of the
population dynamics of the inhabitants of medieval Japan. Comparison of the demographic data of the Yuigahama-minami sample with those of other periods
strongly suggests that the life span of the Yuigahamaminami population was similar to that of the MesolithicNeolithic Jomon population, but was half that of the
early modern Edo population. This finding helped us
detect a secular trend in the life expectancy of inhabitants of Japan over the past several thousands of years.
However, this study encountered problems in terms of
demographic methodology. As previously pointed out
(Bocquet-Appel and Masset, 1982, 1985, 1996; Walker
et al., 1998; Mensforth, 1990; Konigsberg and Frankensberg, 1994), infant underrepresentation and systematic
underestimation of age in adults are unavoidable in
paleodemography. In spite of the difficulties in demographic methodology, the analyses of the Siler hazard
model and of the juvenility index (Bocquet-Appel and
Masset, 1996) allow us to estimate demographic parameters without the bias due to infant underrepresentation
in osteological collections, and also to control for systematic bias in the calculation of adult age distributions.
More importantly, the purpose of this study was to
compare newly obtained medieval data with the data
compiled by Kobayashi (1967), and to discuss the secular
trend in Japanese life expectancy. The methodological
limitations of infant underrepresentation and systematic
underestimation of age in adults is common to both our
data and to the data of Kobayashi (1967). If the assumptions of demographic methodology proposed above (e.g.,
the assumptions of burial, excavation, and skeletal preservation conditions) can be applied to the cases studied
by Kobayashi (1967), a comparison of data from different
periods provides us with a great deal of information
regarding chronological differences in Japanese population structures and life span. This study was an investigation to address the demographic features of the medieval population in Japan, and to refine ideas of a longterm trend in demographic profiles of the inhabitants of
Japan. The results of this study will afford new perspectives on paleodemography in Japan.
LITERATURE CITED
Alesan A, Malgosa A, Simó C. 1999. Looking into the demography of an Iron Age population in the western Mediterranean.
I. Mortality. Am J Phys Anthropol 110:285–301.
Angel JL. 1947. The length of life in ancient Greece. J Gerontol
2:18–24.
Angel JL. 1954. Human biology, health and history in Greece
from the first settlement until now. Yrbk Am Philos Soc 98:
168–174.
Angel JL. 1969. The basis of paleodemography. Am J Phys Anthropol 30:427–438.
Black TK III. 1978. A new method for assessing the sex of fragmentary skeletal remains: femoral shaft circumference Am J Phys
Anthropol 48:227–231.
Bocquet-Appell JP, Masset C. 1982. Farewell to paleodemography. J Hum Evol 11:321–333.
Bocquet-Appell JP, Masset C. 1985. Paleodemography: resurrection or ghost? J Hum Evol 14:107–111.
Bocquet-Appell JP, Masset C. 1996. Paleodemography: expectancy
and false hope. Am J Phys Anthropol 99:571–583.
Brothwell DR. 1981. Digging up bones. Ithaca: Cornell University
Press.
Bruzek J. 2002. A method for visual determination of sex using
the human hip bone. Am J Phys Anthropol 117:157–168.
Charles AK. 1999. First rib metamorphosis: its possible utility
for human age-at-death estimation. Am J Phys Anthropol 110:
303–323.
Drusini AG, Carrara N, Orefici G, Rippa-Bonati M. 2001. Paleodemography of the Nasca Valley populations: reconstruction
of the ancient human ecology in the southern Peruvian coast.
Homo 52:157–172.
Erickson MF. 1978. Aging in the lumbar spine. Am J Phys
Anthropol 48:241–246.
Ferembach D, Schwidetzky I, Stloukal M. 1980. Recommendation
for age and sex diagnoses of skeletons (Workshop of European
Anthropologists). J Hum Evol 9:517–549.
Flecker H. 1942. Time of appearance and fusion of ossification centers as observed by roentgenographic methods. AJR 47:97–159.
Fukawa T. 2002. Mortality event and life table. In: Editorial
Board of the Population Association of Japan, editor. Encyclopedia of population. Tokyo: Baifukan. p 518–522.
PALEODEMOGRAPHY OF MEDIEVAL POPULATION IN JAPAN
Gage TB. 1988. Mathematical hazard models of moratality: an alternative to model life tables. Am J Phys Anthropol 76:429–441.
Gage TB. 1989. Bio-mathematical approach approaches to the
study of human variation and mortality. Yrbk Phys Anthropol
32:184–214.
Gage TB. 1990. Variation and classification of human age pattern of mortality: analysis using competing hazard models.
Hum Biol 62:589–617.
Gage TB, Dyke B. 1986. Parameterizing abridged mortality tables:
the Siler three-component hazard model. Hum Biol 58:275–291.
Gage TB, Mode C. 1993. Some laws of mortality: how well do
they fit? Hum Biol 65:445–461.
Genovés S. 1959. L’estimation des differences sexuelles dans l’os
coxal: differences metriques et differences morphologiques. Bull
Mem Soc Anthropol Paris 10:3–95.
Gordon CG, Buikstra JE. 1981. Soil pH, bone preservation, and
sampling bias at mortuary sites. Am Antiq 46:566–571.
Hanihara K. 1952. Age change in the male Japanese pubic
bone. J Anthropol Soc Nippon 62:245–260.
Hanley BS. 1987. Urban sanitation in preindustrial Japan. J Interdiscipl Hist 18:1–26.
Herrmann NP, Konigsberg LW. 2002. A re-examination of the
age-at-death deistribution of Indian Knoll. In: Hoppa RT, Vaupal
JW, editors. Paleodemography. Cambridge: Cambridge University Press. p 243–257.
Hirata K, Nagaoka T. No date. On human skeletal remains from
Chusei Shudan Bochi site (no. 372). Kamakura: Yuigahama
Chusei Shudan Bochi Iseki Hakkutsu Chosa Dan.
Hirata K, Oku C, Hoshino K, Tomo S, Takahashi S. 2002. On
human skeletal remains from Yuigahama-minami site. In:
Yuigahama-minami Iseki Hakkutsu Chosa Dan, editor. Yuigahama-minami Iseki 2. Kamakura: Yuigahama-minami Iseki
Hakkutsu Chosa Dan. p 1–240.
Hirata K, Nagaoka T, Hoshino K. 2004. Analysis of injuries by
swords in medieval Japanese skeletons from Yuigahama, Kamakura. Anthropol Sci (J Ser) 112:19–26.
Holman DJ. 2003. mle: programming language for building likelihood models. Version 2.1. Website: http://faculty.washington.
edu/djholman/mle/.
Houghton P. 1974. The relationship of the pre-auricular groove
of the ilium to pregnancy. Am J Phys Anthropol 41:381–390.
Howell N. 1979. The demography of the Dobe! Kung. New York:
Academic Press. p 73–97.
_ ˛can MY, Yoshono M, Kato S. 1994. Sex determination from the tibia:
Is
standards for contemporary Japan. J Forensic Sci 39:785–792.
Kawano S. 1989. On population size estimated from archaeological remains. In: Ishii S, Omiwa T, editors. Yomigaeru Chusei
3. Tokyo: Heibonsha. p 61–63.
Kawano S. 1995. On medieval city Kamakura. Tokyo: Kodansha.
Kito H. 2000. On demographic perspective of Japanese history.
Tokyo: Kodansha.
Kobayashi K. 1967. Trend in the length of life based on human
skeleton from prehistoric to modern times in Japan. J Fac Sci
Univ Tokyo Sec V 3:107–162.
Konigsberg LW, Frankensberg SR. 1994. The paleodemography:
‘‘not quite dead.’’ Evol Anthropol 3:92–105.
Krogman WH. 1962. The human skeleton in forensic medicine.
Springfield, IL: Charles C. Thomas.
Lovejoy CO. 1971. Methods for detection of census error. Am
Anthropol 73:101–109.
Lovejoy CO. 1985. Dental wear in the Libben population: its
functional pattern and role in the determination of adult skeletal age at death. Am J Phys Anthropol 68:47–56.
Lovejoy CO, Meindl RS, Pryzbeck TR, Barton TS, Kotting D,
Heiple KG. 1977. The palaeodemography of the Libben site,
Ottowa County, Ohio. Science 198:291–293.
Lovejoy CO, Meindl RS, Pryzbeck TR, Mensforth RP. 1985.
Chronological metamorphosis of the auricular surface of the
illium: a new method of determining adult age at death. Am
J Phys Anthropol 68:15–28.
Makeham WM. 1860. On the law of mortality. J Inst Actuar 13:
325–358.
Martin R, Knussmann R. 1988. Anthropologie. Band 1. Stuttgart:
Gustav Fischer.
13
Matsushita T. 2002. On medieval human skeletal remains from
Yuigahama-minami site. In: Yuigahama-minami Iseki Hakkutsu
Chosa Dan, editor. Yuigahama-minami Iseki 3. Kamakura:
Yuigahama-minami Iseki Hakkutsu Chosa Dan. p 1–134.
Mays S. 1998. The archaeology of human bones. London: Routledge.
Meindl RS, Lovejoy CO. 1985. Ectocranial suture closure: a revised
method for the determination of skeletal age at death on the lateral-anterior sutures. Am J Phys Anthropol 68:57–66.
Mensforth BP. 1990. Paleodemography of the Carlston Annis
(Bt-5) late archaic skeletal population. Am J Phys Anthropol
82:81–99.
Mensforth RP, Lovejoy CO. 1985. Anatomical, physiological, and
epidemiological correlates of the aging process: a confirmation
of multifactorial age determination in the Libben skeletal population. Am J Phys Anthropol 68:87–106.
Miles AEW. 1963. The dentition in the assessment of individual
age in skeletal material. In: Brothwell DR, editor. Dental anthropology. Oxford: Pergamon. p 191–209.
Milner GR, Wood JW, Boldsen JL. 2000. Paleodemography. In:
Katzenberg MA, Saunders SR, editors. Biological anthropology of the human skeleton. New York: Wiley-Liss. p 467–
497.
Ministry of Health, Labor and Welfare. 2004. Abridged life table
in 2003. Tokyo: Ministry of Health, Labor and Welfare.
Mizushima H. 1961. On prefecture-specific life tables from 1921 to
1956. Tokyo: Zaidan Hojin Seimei Hoken Bunka Kenkyusho.
Morimoto I, Takahashi Y, Hirata K. 1984. On human skeletons.
In: Yuigahama Chusei Shudan Bochi Iseki Hakkutsu Chosa
Dan, editor. Yuigahama Chusei Shudan Bochi Iseki Hakkutsu
Chosa Houkokusho. Kamakura: Educational Committee of
Kamakura City. p 8–11.
Nagaoka T, Hirata K. 2006. Tooth size of the medieval period
people of Japan. Anthropol Sci DOI: IO.1537/ase.04111
(in press).
Nagar Y, Hershkovitz I. 2004. Interrelationship between various
aging methods, and their relevance to palaeodemography. Hum
Evol 19:145–156.
Nakahashi T, Nagai M. 1985. On human skeletons from Yoshimohama site. In: Shimonoseki-shi Kyoiku Iinkai, editor. Yoshimohama
Iseki. Shimonoseki: Educational Committee of Shimonoseki City.
p 154–225.
Nakahashi T, Nagai M. 1989. Yayoi people. In: Nagai M, Nasu T,
Kanaseki T, Sahara M, editors. Yayoi-bunka no Kenkyu 1. Tokyo:
Yuzankaku. p 76–95.
Nathan H. 1962. Osteophytes of the vertebral column. An anatomical study of their development according to age, race and
sex with considerations as to their etiology and significance.
J Bone Joint Surg [Am] 44:243–268.
Okada M. 1961. On the age-related changes of cranial sutures
in Kanto Japanese. Jikei Med J 77:112–167.
Phenice TW. 1969. A newly developed visual method of sexing
the os pubis. Am J Phys Anthropol 30:297–301.
Safont S, Malgosa A, Subirà ME. 2000. Sex assessment on the
basis of long bone. Am J Phys Anthropol 113:317–328.
Sakai S. 1982. On the history of medicine. Tokyo: Tokyo Shoseki.
Siler W. 1979. A competing-risk model for animal mortality.
Ecology 60:750–757.
Siler W. 1983. Parameters of mortality in human populations
with widely varying life span. Stat Med 2:373–380.
SPSS, Inc. 2005. SPSS for Windows. Version 13.0J. Chicago:
SPSS, Inc.
Stewart TD. 1958. The rate of development of vertebral osteoarthritis in American whites and its significance in skeletal
age identification. Leech 28:144–151.
Suzuki H, Watanabe H, Iwamoto M, Masuda S, Inamoto N,
Hayashi T, Tanabe Y, Sakura H, Kohara Y. 1956. Medieval Japanese skeletons from the burial site at Zaimokuza, Kamakura
City. In: Anthropological Society of Nippon, editor. Kamakurashi Zaimokuza Hakken no Chusei Iseki to sono Jinkotsu. Tokyo:
Iwanami Shoten. p 1–74.
Suzuki T. 1998. On information from human skeletal remains.
In: Baba H, editor. Kokogaku to Jinruigaku. Tokyo: Doseisha.
p 83–120.
14
T. NAGAOKA ET AL.
Ubelaker DH. 1989. Human skeletal remains. Excavation, analysis, interpretation, 2nd ed. Chicago: Aldine.
Wakebe T. 1990. A morphological study of cranium of infants
and children in the Japanese. Nagasaki Med J 65:805–824.
Walker PL, Johnson JR, Lambert PM. 1998. Age and sex biases
in the preservation of human skeletal remains. Am J Phys
Anthropol 76:183–188.
Webb PAO, Suchey JM. 1985. Epiphyseal union of anterior iliac
crest and medial clavicle in a modern multiracial sample of
American males and females. Am J Phys Anthropol 68:457–
466.
Weiss MW. 1973. Demographic models for anthropology. Memoirs
of the Society for American Archaeology 27. Am Antiq 38:1–186.
White TD, Folkens PA. 2000. Human osteology, 2nd ed. London:
Academic Press.
Wood JW, Holman DJ, Weiss KM, Bucanan AV, LeFor B. 1992.
Hazards models for human population biology. Yrbk Phys
Anthropol 35:43–87.
Wood JW, Holman DJ, O’Connor KA, Ferrell RJ. 2002. Mortality models for paleodemography. In: Hoppa RT, Vaupal JW,
editors. Paleodemography. Cambridge: Cambridge University
Press. p 129–168.