Original Articles
Preauricular grooves on Japanese female skeletal remains in Edo as a bioarchaeological indicator in population studies
2025 Volume 133 Issue 1 Pages 1-15
Details
2025 Volume 133 Issue 1 Pages 1-15
Preauricular grooves in women are considered as an indicator of childbirth or pregnancy. We examined 522 female pelvises excavated from archaeological sites in Edo (present-day Tokyo), Japan, which underwent population growth during the Edo period (AD 1603–1867). We analyzed the frequency of preauricular grooves by estimated age group, burial period, and type of burial facility in Edo to infer age-related changes and chronological as well as social-rank-related differences during the Edo period. We found that the frequency of preauricular grooves increased with individual age. Moreover, a higher frequency of developed preauricular grooves was observed during and after the 18th century (the middle–late Edo period) compared with those observed in the 17th century (early Edo period) in all age groups. The frequencies of preauricular grooves in remains buried in kamekan (ceramic coffins) and hayaoke (circular wooden coffins) revealed no social differentiation of reproduction rate. These results suggest that the average number of births per woman during and after the 18th century may have exceeded that in the 17th century. Consequently, it is highly probable that the population increase known to occur in Edo during the 17th century was caused by an influx from rural areas rather than by an increasing number of births among the city’s female residents. Furthermore, the number of pregnancies and births did not differ significantly among different social classes during the later Edo period.
The preauricular groove is a sulcus or pit-like bony depression that forms differentially on the lower anterior part of the iliac preauricular surface in both sexes (Derry, 1911; Hoshi, 1961; Houghton, 1974, 1975; Buikstra and Ubelaker, 1994; Bruzek, 2002; Steckel et al., 2006; Maass and Friedling, 2016). Those found specifically in women are referred to as ‘pregnancy scars’ or ‘parturition scars’ because they are generally believed to develop during pregnancy and childbirth.
Previous studies have explored the relationship between the number of pregnancies and births experienced by women and the degree of development of preauricular grooves, indicating the history of pregnancy and childbirth, using female skeletal materials (Dunlap, 1979; Kelley, 1979; Işcan and Dunlap, 1984; Igarashi, 1992a; Schemmer et al., 1995; Igarashi et al., 2020).
Schemmer et al. (1995) applied radiography to examine the preauricular grooves in 70 female skeletal samples and found that more developed grooves were common in multiparous women, whereas they were absent in nulliparous women. Igarashi et al. (2020) delineated three stages of development of preauricular grooves. Their examination of 90 donated female cadavers revealed that the least developed stage only occurred in nulliparous women, whereas the most advanced stage generally occurred in multiparous women. However, several studies have refuted the existence of any strong association between the preauricular groove and the number of pregnancies and births (e.g. Spring et al., 1989; Cox and Scott, 1992). In response, Igarashi et al. (2020) argued that such negative findings can be attributed to differences in the methods used to classify preauricular grooves, the accuracy of available information on reproductive history, or the statistical methods used in the analyses.
In light of the finding that the formation of a preauricular groove reflects a woman’s reproductive history, researchers have sought to estimate aspects of population reproduction using the rate of occurrence and development of preauricular grooves in female skeletal remains as an indicator of fertility (Ullrich, 1975; Phillipps, 1980; Igarashi, 1992b, 2022; Igarashi et al., 2020). Within the Japanese population, Igarashi (1992b) examined preauricular groove formation in the Jomon, a prehistoric Japanese population. She found that the development of the preauricular groove increased with age in this population and also differed across regional subgroups, with the Jomon in Hokkaido (the northernmost of Japan’s four main islands) exhibiting more developed grooves than other regional subgroups.
Although studies have been conducted on prehistoric archaeological populations in the Japanese archipelago (Igarashi, 1992b, 2022; Igarashi et al., 2020), no such systematic analysis has been conducted for skeletal remains dated to historic periods for which written records are available. To the best of our knowledge, this study marks the first attempt to investigate the frequency of occurrence of preauricular grooves during the Edo period (AD 1603–1867), which has been a topic of research pertaining to fertility and demographics within various fields of historical science. The population of Japan in 1600 has been estimated to have been between twelve million (Hayami, 1975) and sixteen million (Farris, 2006; Saito and Takashima, 2015), but the drastic increase in the marriage rate following the dissolution of the extended family system is known to have increased the population to over thirty million by 1721, suggesting that the 17th century was a period of population explosion. Subsequently, it is known that population growth stagnated in the 18th century and slightly regrew in the 19th century (Kito, 1996).
The analysis was conducted using human remains excavated in the central area of Tokyo, formerly the city of Edo. In recent years, the number of available human skeletons has increased dramatically because of urgent excavations accompanying developmental projects.
Following the establishment of the Tokugawa Shogunate in 1603, the population of Edo, which was just an ordinary castle town in the medieval period, grew rapidly throughout the 17th century and is estimated to have exceeded one million by the first quarter of the 18th century (Kito, 1989).
In Edo, the Tokugawa Shogunate began censuses of the townspeople every six years from 1721, and records of these censuses are still available, showing temporal changes in the population and sex ratio. However, there are no records of the population of townspeople before the census began, or of the population of samurai and priests throughout the Edo period. Historical studies have assumed that the population grew rapidly in the early Edo period due to the influx of people from neighboring rural areas, and that the population stabilized around one million in the late Edo period due to the balance between the inflow and birth rate, and the outflow and death rate (Sekiyama, 1958; Minami, 1978; Saito, 2002; Hayami, 2008).
Some studies of documentary history attribute this rapid population increase to the inflow of laborers due to urban development after the start of large-scale development in Edo, such as the widening of Edo Castle and the reclamation of the inlet near the castle (Kito, 1989, 2010). The institutionalization of the ‘Sankin-kotai’ system, in which daimyos (feudal lords) from all over Japan were required to live in Edo periodically and their family members settled in Edo with many of their vassals, is also thought to have contributed to the increased number of immigrants (Maruyama, 2007).
Kito (1989) also noted that the population of townspeople in Edo, which was extremely male-dominated in the 17th century, experienced an increase in the proportion of female population in the second half of the 18th century. The female population, which was only 55.1% male in 1721 and 59.0% in 1747, reached 83.4% in 1832, after a period of 85 years without records of the sex ratio, and 97.9% in 1867, the last year of the Edo period (Koda, 1938). This showed a trend toward an equivalent sex ratio over time in Edo.
Despite this accumulation of research on population change, most of the documents similar to contemporary family registers that could have been used to estimate the number of births in Edo were lost in air raids during World War II. Therefore, we have to rely on bioarchaeological methods to search for birth rates and natural increase rates.
Tsutaya et al. (2014), who estimated the duration of the weaning process according to the nitrogen isotope ratios of 84 subadult skeletons excavated from the Hitotsubashi High School site, inferred that Edo’s crude birth rate in the 17th century would not have been high. They estimated that at the end of the weaning process, a child would be approximately 3.1 years (95% confidence interval, 2.1–4.1 years), indicating a relatively long lactation period. It is known that during lactation, the resumption of maternal ovulation is suppressed by changes in endocrine hormone dynamics and that later weaning leads to longer birth intervals (Tsutaya, 2018). The combination of a long lactation period and an imbalance in the sex ratio would have limited the number of births in Edo in the 17th century. Coupled with the tendency for the people at this site to have short life expectancy (Nagaoka and Hirata, 2007), these sociodemographic conditions likely impeded the natural population increase, aligning with historical assumptions that the urbanization of Edo in the early Edo period was supported by in-migration (Tsutaya et al., 2014).
On the other hand, Tsutaya et al. (2019) performed nitrogen isotope analysis on 40 subadult skeletons excavated from the Sakai Kango Toshi 871 site in Osaka, Japan, estimated to have been buried between the late 17th and 19th centuries, later than the burial period at the Hitotsubashi High School site. They estimated that at the end of the weaning process, a child would be approximately 1.9 years (95% confidence interval, 1.4–2.7 years), indicating that the lactation period was shorter than that of the Hitotsubashi High School site population. They attributed this difference in the end of the weaning age to the population reproduction strategy of shortening the lactation period and the birth interval to ensure legitimate children, as the inherited family consciousness spread to the common people after the late 17th century.
Although estimates of the end of the weaning process may be related to the potential number of children a woman can produce, it is quite possible that it does not directly correlate with the actual number of births for people in the Edo period, who often adopted reproductive strategies that deliberately curbed the number of births (Sawayama, 2017). In addition, as noted by Tsutaya et al. (2019), the differences in the end of the weaning process discussed above may reflect regional differences rather than temporal differences between the two sites.
Therefore, when discussing temporal differences in fertility in Edo, it would be useful to compare the frequency of occurrence of preauricular grooves, which can be interpreted as traces of more direct childbearing behavior, among skeletons dated to different periods excavated in Edo. Thus, we compared the frequency of preauricular grooves observed in remains buried during the period of population growth and those buried after this period.
The archaeological design of Edo also reflects residents’ social status. The burial facilities in the city’s cemeteries generally correspond to the social status of the deceased. Those buried in wooden circular coffins were mostly ordinary townspeople, whereas those buried in ceramic coffins belonged to the higher-ranked samurai class (Tanigawa, 1987, 2004; Taguchi, 2009). Therefore, the status of the deceased people can be inferred from the type of burial facility. We examined the effect of social stratification on fertility and reproductive strategies in Edo by comparing the frequency of preauricular grooves on the skeletal remains of female pelvises found in ceramic and circular wooden coffins.
The materials examined were bones from female skeletal remains excavated from archaeological sites of temples located in Edo. The following selection criterion was applied: at least one of the two ilia could be observed to determine the type of preauricular groove, as described below. Some of the specimens obtained had morphological anomalies induced by diseases such as arthritis, periostitis, syphilis, and osteoporosis. We did not exclude such individuals from our study because they were considered to be representative of people who lived during the Edo period as well as those with no bone lesions. Although we mostly used materials with the lower part of the auricular surface intact, a small sample of materials exhibiting damage to the lower edge of the auricular surface was also included. In the latter, the extent of damage was limited to less than one-fifth of the lower part of the auricular surface; those cases evidencing more damage were excluded. Skeletal remains of 522 individuals across 15 sites were used in this study (Table 1, Figure 1).
Skeletal materials by age group obtained at different sites and estimated dates of graveyard establishment
| Site | Graveyard formation period | Estimated age | Total | ||||
|---|---|---|---|---|---|---|---|
| Adolescent | Young | Middle | Old | Unknown | |||
| Genkoin site1 | 17th–early 18th century | 2 | 2 | 7 | 1 | 3 | 15 |
| Kounji site2 | 1695–Meiji era | 3 | 16 | 28 | 11 | 11 | 69 |
| Asakusa-Takaharacho site3 | 1635–1928 | 2 | 5 | 3 | 1 | 3 | 14 |
| Ikenohata-Shichikencho site3 | 1627–1913 | 14 | 51 | 52 | 13 | 4 | 134 |
| Ushigomejo-ato site3 | 1792– | 2 | 1 | 3 | |||
| Unkoin site3 | 1682–Taisho era | 5 | 5 | 12 | 1 | 1 | 24 |
| Gokokuji-Monzenmachi site3 | Late 17th century–1697 | 1 | 1 | 1 | 1 | 4 | |
| Shitaya-Dohocho site3 | Early 16th century–1683 | 1 | 2 | 2 | 1 | 2 | 8 |
| Sugenji site3 | 1653–1899 | 5 | 43 | 33 | 5 | 1 | 87 |
| Shokenji site3 | 1607–1909 | 7 | 18 | 25 | 6 | 5 | 61 |
| Hatchobori-Sanchome site3 | Late 16th century–1657 | 2 | 6 | 15 | 3 | 3 | 29 |
| Hokoji site3 | 1669–1889 | 2 | 4 | 4 | 2 | 3 | 15 |
| Hoshoji site3 | Early 17th century–1910s | 3 | 2 | 1 | 6 | ||
| Hosshoji site3 | 1667–1904 | 11 | 10 | 6 | 3 | 30 | |
| Yanaka-Sansakicho site3 | End of 17th century–Meiji era | 8 | 7 | 4 | 4 | 23 | |
| Total | 44 | 174 | 201 | 57 | 46 | 522 | |
1Stored at Keio University. 2Stored at the Institute of Physical Anthropology, Niigata University of Health and Welfare. 3Stored at the Department of Anthropology, National Museum of Nature and Science.
Locations of temple sites covered in this study.
The burial periods were divided into three categories: the early Edo period, the middle–late Edo period, and unknown (Table 2). We considered the onset of the 18th century as the boundary demarcating the early Edo period and the middle–late Edo period. Burials at the Shitaya-Dohocho site (Katoh Construction Co., Ltd., 2010), the Hatchobori-Sanchome site (Hatchobori sanchome iseki (dainiji) chosakai, 2003), and the Gokokuji-Monzenmachi site (TEIKEI Trade Co., Ltd., Maizo bunkazai jigyobu, 2009) were all categorized under the early Edo period. These temples that once stood on those sites were abandoned or relocated, and the graveyard was established during the 17th century. The burials at the Kounji site (Pasco Corporation, 2021), which were relocated to the present site in 1695, and those at the Ushigomejo-ato site (Tochigi, 2004), which were relocated in 1792, all belonged to the middle–late Edo period. The Unkoin site (Taisei Engineering Co., Ltd., 2010) was moved to the current site in 1682, but the burials were all classified as the middle–late Edo period because the excavated area was a graveyard in which many townspeople were buried during the 18th century. Similarly, all burial facilities at the Asakusa-Takaharacho site (Taisei Engineering Co., Ltd., 2015) found in the excavated area were constructed during or after the mid-18th century and were therefore categorized under the middle–late Edo period. Burial dates for the following sites were determined according to the stratigraphic level of the unearthed burial facilities: the Sugenji and Shokenji sites (Taisei Engineering Co., Ltd., 2005), the Hosshoji site (2nd survey) (Shinjuku-ku Minamimotomachi iseki chosakai, 1991), the Hokoji site (Okasanlivic Co., Ltd., 2008), the Hoshoji site (Taisei Engineering Co., Ltd., 2007), and the Yanaka-Sansakicho site (Taito-ku Bunkazai Chosakai, 2000). A chronological typology was developed following Koremura (2005) to date the kamekan (ceramic coffins) (Ogiura, 1987; Koremura, 2005). Burials of Type A were dated to the early Edo period, while those of Types B to G were dated to the middle–late Edo period. Burial facilities at the Ikenohata-Shichikencho site (Taito-ku Ikenohata shichikencho iseki chosadan, 1997), dated between the 17th century and the early 18th century, as described in the excavation report, were assigned to the early Edo period. Those at the Genkoin site (Minato-ku Shibakoen 1-chome iseki chosadan, 1988), in which only Chinese coins, old Kan’ei Tsuho coins, and Mon Kan’ei Tsuho coins were placed were also assigned to the early Edo period. However, those in which new Kan’ei Tsuho coins were placed were assigned to the middle–late Edo period, according to the year of their first casting. The period of burials at the Hosshoji site (first survey) (Shinjuku-ku Hosshoji ato iseki chosakai, 1991) is unknown because it was difficult to excavate this area stratigraphically. Therefore, only burials in the square wooden coffins, which began to be used in the early 18th century, were assigned to the middle–late Edo period; burials in other kinds of facilities were categorized as ‘unknown.’
Number of burial facilities by type and burial period at each site
| Site | Burial period | Burial type | Total | ||
|---|---|---|---|---|---|
| Kamekan | Hayaokea | Other | |||
| Genkoin site | Early | 14 | 14 | ||
| Middle–late | 1 | 1 | |||
| Kounji site | Middle–late | 43 | 8 | 18 | 69 |
| Asakusa-Takaharacho site | Middle–late | 9 | 4 | 1 | 14 |
| Ikenohata-Shichikencho site | Early | 22 | 1 | 23 | |
| Middle–late | 17 | 59 | 34 | 110 | |
| Unknown | 1 | 1 | |||
| Ushigomejo-ato site | Middle–late | 3 | 3 | ||
| Unkoin site | Middle–late | 22 | 2 | 24 | |
| Gokokuji-Monzenmachi site | Early | 4 | 4 | ||
| Shitaya-Dohocho site | Early | 7 | 1 | 8 | |
| Sugenji site | Middle–late | 18 | 19 | 46 | 83 |
| Unknown | 4 | 4 | |||
| Shokenji site | Early | 2 | 2 | ||
| Middle–late | 18 | 12 | 25 | 55 | |
| Unknown | 4 | 4 | |||
| Hatchobori-Sanchome site | Early | 14 | 15 | 29 | |
| Hokoji site | Early | 5 | 5 | ||
| Middle–late | 6 | 3 | 9 | ||
| Unknown | 1 | 1 | |||
| Hoshoji site | Early | 1 | 1 | ||
| Middle–late | 3 | 2 | 5 | ||
| Hosshoji site | Middle–late | 2 | 9 | 11 | |
| Unknown | 6 | 13 | 19 | ||
| Yanaka-Sansakicho site | Early | 1 | 1 | ||
| Middle–late | 17 | 2 | 3 | 22 | |
| Total | 144 | 208 | 170 | 522 | |
a Burials excavated from hayaokes (circular wooden coffins) belonging to the early Edo period were excluded from the hayaoke group.
Research on burial customs in Edo during the early modern period (e.g. Tanigawa, 2004) has revealed a transition in burial customs. In the 17th century, hayaoke (circular wooden coffins) were the predominant burial facility used throughout Edo (Nakayama, 2017), and were assumed to be used even for those belonging to middle- or high-ranked samurai classes, including the shoguns (the supreme authorities in the feudal state) (Nara, 1988). During and after the 18th century, some degree of correspondence between the type of burial facility and social status was evident. The kamekan, surrounded by wooden chambers, were used mainly for middle-ranking and high-ranking samurai, who occupied political positions and enjoyed high levels of prestige and privilege. Kamekan without chambers are believed to have been used mainly for low-ranking samurai. By contrast, hayaoke were basically used for townspeople, including merchants and artisans, as well as for a small section of lower-ranking samurai (Tanigawa, 2004; Taguchi, 2009, 2012).
Considering the above information, we defined burials entailing ceramic coffins, with or without wooden chambers, as the kamekan group and post-18th-century burials in circular wooden coffins as the hayaoke group. This delineation facilitates our analysis of the degree of development of preauricular grooves in groups belonging to different social classes. Square wooden coffins, which appeared during the first half of the 18th century and increased from the mid-18th century, were not considered in this study because the status of these burials remains unclear (Ikumi, 1991; Kuramochi and Suzuki, 2003; Tanigawa, 2004; Suzuki and Noshiro, 2006; Yonemoto, 2012).
Sex and ageSex was determined on the basis of pelvic morphology (Phenice, 1969; Bruzek, 2002) and cranial morphology (Buikstra and Ubelaker, 1994), and only female pelvises were considered in this study. Four age groups (at the time of death) were delineated: ‘adolescent,’ ‘young,’ ‘middle,’ and ‘old’ (Table 3) according to the degree of ossification of the epiphyses (Schaefer et al., 2009) and metamorphosis of the iliac auricular surface (Buckberry and Chamberlain, 2002), and of the pubic symphysis (Brooks and Suchey, 1990; Sakaue, 2006). Pelvises that did not exhibit acetabular ossification in the os coxae were excluded from the study because they were presumed to have not yet reached reproductive maturity (Seta and Yoshino, 1990; Schaefer et al., 2009).
Criteria used to estimate age groups
| Estimated age stages | Status of epiphyses | Auricular surface | Pubic symphysis |
|---|---|---|---|
| Adolescent | One or more long bone ends have not fused or are in the process of fusion, or the either or both clavicle sternal ends or iliac crests have not fused | Stage 1 | Phase 1–2 |
| Young | Either or both clavicle sternal ends or iliac crests are in the process of fusion | Stage 2 | Phase 3 |
| Middle | — | Stage 3–4 | Phase 4–5 |
| Old | — | Stage 5–7 | Phase 6 |
Age groups were estimated according to the status of the epiphyses, auricular surface, and pubic symphysis in descending order of priority.
Following Igarashi et al. (2020), three types of preauricular groove were distinguished according to their appearance: ‘no change,’ ‘weak,’ and ‘strong.’
We delineated three types under the ‘no change’ category. In the first, the bone surface was smooth and showed no change (Type 1 in Igarashi et al., 2020). The second type evidenced some palpable depressions but no observable change (Type 2 in Igarashi et al., 2020). The third type, which may not have been related to pregnancy or parturition, showed observable depressions in both female and male pelvises (Type 3 in Igarashi et al., 2020).
Both the ‘weak’ and ‘strong’ types of preauricular grooves had true depressions with fully closed contours, which did not run parallel to the ventral edge of the auricular surface (Igarashi et al., 2020). These features are only found in female innominate bones, occurring as pregnancy and parturition scars. Weak preauricular grooves had singular contours resembling a groove or coalescence of a series of pits (Type 4 in Igarashi et al., 2020); strong ones exhibited a true depression with double contours, namely, inner contours enveloped by the outer contour (Type 5 in Igarashi et al., 2020). According to Igarashi et al. (2020), the strong groove type could form as a result of the superimposition of several scars. Figure 2 shows typical examples of each of these preauricular groove types.
Classification of preauricular grooves: (a) an example of a ‘no change’ type of preauricular groove, in which the contour of the depression is unclear; (b) an example of a ‘weak’ type evidencing two independent grooves with completely closed contours (indicated by black arrows), which do not run parallel to the ventral edge of the auricular surface; (c) an example of a ‘strong’ type with closed inner contours (indicated by white arrows) within the closed outer contour (indicated by the black arrow). We followed the classification method proposed by Igarashi et al. (2020).
In this study we analyzed the prevalence of the above three preauricular groove types (‘no change,’ ‘weak,’ and ‘strong’), as well as three preauricular groove classes (TS0 = total score 0; TSL = total score lower; TSH = total score higher; see Table 4), as defined by the scoring system of Igarashi et al. (2020). For remains in which only one side of the preauricular groove was visible, the preauricular groove class was determined following the assumption that the opposite side had the same preauricular groove type.
Correspondences between the type and class of preauricular groove
| Right preauricular groove type | Left preauricular groove type | Preauricular groove classa |
|---|---|---|
| No change | No change | TS0 |
| No change | Weak | TSL |
| No change | Strong | TSL |
| Weak | No change | TSL |
| Weak | Weak | TSL |
| Weak | Strong | TSH |
| Strong | No change | TSL |
| Strong | Weak | TSH |
| Strong | Strong | TSH |
a Proposed by Igarashi et al. (2020). Abbreviations: TS0 = total score 0; TSL = total score lower; TSH = total score higher.
Igarashi et al. (2020) quantitatively demonstrated that the preauricular groove class is related to the number of pregnancies and births. It is therefore an indispensable index for exploring how preauricular groove development is linked to the reproductive history of an individual or population. However, the class of the preauricular groove is also an index that discounts the presence or absence of left–right variations, as it only considers combined left and right preauricular groove types. In addition, there are many cases in which one side of the preauricular groove is missing in an excavated human skeletal specimen. Therefore, limiting the samples to skeletal remains in which both sides of preauricular grooves are present would greatly reduce the data available for analysis. To alleviate these issues, we analyzed not only the preauricular groove class but also the left and right preauricular groove types to ascertain whether the same trend emerged for both indices.
Data analysisWe compared the frequencies of the ‘no change’ and ‘strong’ types of preauricular grooves and the TS0 and TSH classes in the different estimated age groups, burial periods, and burial facilities. Fisher’s exact test was used to determine independence at the 5% level of significance. For multiple comparisons, we applied the Hochberg method to correct for the probability of significance.
We also compared the frequencies of the TS0 and TSH classes between different burial periods or burial facilities within the estimated age groups to see if the observed differences in overall preauricular groove class frequency can be attributed to differences in the age structure of each group.
All analyses and plots were generated in R version 4.3.0. (R Core Team, 2023).
We found that the frequencies of occurrence of the ‘strong’ type and TSH class increased with age, whereas those for the ‘no change’ type and TS0 class decreased (Table 5, Table 6; Figure 3, Figure 4). In the ‘adolescent’ group, more than half of the materials exhibited the ‘no change’ type of preauricular groove, and no ‘strong’ type was observed.
Frequency of observations of the different types of preauricular grooves (‘no change,’ ‘weak,’ and ‘strong’) in the estimated age groups
| No change | Weak | Strong | Total | ||
|---|---|---|---|---|---|
| Adolescent | R | 22 (53.7%) | 19 (46.3%) | 0 (0.0%) | 41 |
| L | 19 (52.8%) | 17 (47.2%) | 0 (0.0%) | 36 | |
| Young | R | 57 (36.1%) | 87 (55.1%) | 14 (8.9%) | 158 |
| L | 53 (35.8%) | 83 (56.1%) | 12 (8.1%) | 148 | |
| Middle | R | 58 (30.7%) | 108 (57.1%) | 23 (12.2%) | 189 |
| L | 45 (25.4%) | 106 (59.9%) | 26 (14.7%) | 177 | |
| Old | R | 8 (15.1%) | 28 (52.8%) | 17 (32.1%) | 53 |
| L | 8 (14.8%) | 30 (55.6%) | 16 (29.6%) | 54 |
Frequency of observations of the preauricular groove classes in the estimated age groups
| TS0 | TSL | TSH | Total | |
|---|---|---|---|---|
| Adolescent | 22 (50.0%) | 22 (50.0%) | 0 (0.0%) | 44 |
| Young | 46 (26.4%) | 107 (61.5%) | 21 (12.1%) | 174 |
| Middle | 46 (22.9%) | 115 (57.2%) | 40 (19.9%) | 201 |
| Old | 7 (12.3%) | 29 (50.9%) | 21 (36.8%) | 57 |
Abbreviations: TS0 = total score 0; TSL = total score lower; TSH = total score higher.
Frequencies of preauricular groove types in the estimated age groups (right and left, in that order).
Frequencies of preauricular groove classes in the estimated age groups.
We examined pairwise differences in frequencies of the ‘no change’ type and the ‘strong’ type on the right and left sides of the grooves, considered separately, for the estimated age groups. Significant differences in the frequencies of the ‘no change’ type of preauricular grooves on both the left and right sides were found between ‘adolescent’ and ‘middle,’ ‘adolescent’ and ‘old,’ and ‘young’ and ‘old.’ We found significant differences in the frequencies of the ‘strong’ type between ‘adolescent’ and ‘middle’ only on the left side and between ‘adolescent’ and ‘old,’ ‘young’ and ‘old,’ and ‘middle’ and ‘old’ on both sides of the grooves (Table 7).
Results of significance tests for preauricular groove types (‘no change’ or ‘strong’) for paired age groups
| Right | Left | |||
|---|---|---|---|---|
| No change | Strong | No change | Strong | |
| Adolescent–young | ||||
| Adolescent–middle | * | * | * | |
| Adolescent–old | ** | ** | ** | ** |
| Young–middle | ||||
| Young–old | * | ** | * | ** |
| Middle–old | ** | * | ||
* P < 0.05, ** P < 0.01. The Hochberg method was used to correct for significance level.
Many of the pairwise differences were also significant for the preauricular groove classes; the frequencies of occurrence of the TS0 class differed significantly between ‘adolescent’ and ‘young,’ ‘adolescent’ and ‘middle,’ and ‘adolescent’ and ‘old.’ The frequencies of TSH classes for all pairwise combinations differed significantly (Table 8).
Results of significance tests for TS0 and TSH in the different age groups
| TS0 | TSH | |
|---|---|---|
| Adolescent-young | * | * |
| Adolescent-middle | ** | ** |
| Adolescent-old | ** | ** |
| Young-middle | * | |
| Young-old | ** | |
| Middle-old | * |
* P < 0.05, ** P < 0.01. The Hochberg method was used to correct for significance level. Abbreviations: TS0 = total score 0; TSH = total score higher.
We found higher frequencies of the developed preauricular groove type and class in the middle–late Edo period compared with the early Edo period (Table 9, Table 10; Figure 5, Figure 6). Fisher’s exact test results revealed a significant difference (P < 0.05) only in the frequency of the ‘strong’ type of groove on the left side. We also found that the percentage of ‘strong’ type of both sides and TSH is higher in the middle–late Edo period than in the early Edo period for all estimated age groups except for ‘adolescent,’ which has no strong type on both sides (Figure 7, Figure 8). In the comparisons between the early and the middle–late Edo periods within each age group, however, we could find a significantly unbalanced distribution only in the old age group in the left side, where the strong type of preauricular groove was found more (35.6%) in the middle–late Edo series than in the early Edo (Table 9).
Frequency of preauricular groove types (‘no change,’ ‘weak,’ and ‘strong’) for different burial periods categorized by age group
| No change | Weak | Strong | Total | |||
|---|---|---|---|---|---|---|
| Early | R | Adolescent | 5 (62.5%) | 3 (37.5%) | 0 (0.0%) | 8 |
| Young | 7 (38.9%) | 11 (61.1%) | 0 (0.0%) | 18 | ||
| Middle | 11 (29.7%) | 22 (59.5%) | 4 (10.8%) | 37 | ||
| Old | 1 (14.3%) | 4 (57.1%) | 2 (28.6%) | 7 | ||
| Unknown | 2 (33.3%) | 4 (66.7%) | 0 (0.0%) | 6 | ||
| L | Adolescent | 4 (50.0%) | 4 (50.0%) | 0 (0.0%) | 8 | |
| Young | 8 (40.0%) | 12 (60.0%) | 0 (0.0%) | 20 | ||
| Middle | 8 (20.5%) | 27 (69.2%) | 4 (10.3%) | 39 | ||
| Old | 0 (0.0%) | 7 (100.0%) | 0 (0.0%) | 7 | ||
| Unknown | 4 (50.0%) | 3 (37.5%) | 1 (12.5%) | 8 | ||
| Middle–late | R | Adolescent | 17 (53.1%) | 15 (46.9%) | 0 (0.0%) | 32 |
| Young | 47 (35.9%) | 72 (55.0%) | 12 (9.2%) | 131 | ||
| Middle | 45 (30.8%) | 84 (57.5%) | 17 (11.6%) | 146 | ||
| Old | 6 (14.0%) | 24 (55.8%) | 13 (30.2%) | 43 | ||
| Unknown | 12 (48.0%) | 11 (44.0%) | 2 (8.0%) | 25 | ||
| L | Adolescent | 15 (55.6%) | 12 (44.4%) | 0 (0.0%) | 27 | |
| Young | 45 (37.2%) | 65 (53.7%) | 11 (9.1%) | 121 | ||
| Middle | 35 (26.5%) | 75 (56.8%) | 22 (16.7%) | 132 | ||
| Old | 8 (17.8%) | 21 (46.7%) | 16 (35.6%) | 45 | ||
| Unknown | 5 (20.8%) | 17 (70.8%) | 2 (8.3%) | 24 |
Frequency of preauricular groove classes for different burial periods categorized by age group
| TS0 | TSL | TSH | Total | ||
|---|---|---|---|---|---|
| Early | Adolescent | 4 (44.4%) | 5 (55.6%) | 0 (0.0%) | 9 |
| Young | 5 (25.0%) | 15 (75.0%) | 0 (0.0%) | 20 | |
| Middle | 8 (20.0%) | 26 (65.0%) | 6 (15.0%) | 40 | |
| Old | 0 (0.0%) | 6 (75.0%) | 2 (25.0%) | 8 | |
| Unknown | 5 (50.0%) | 4 (40.0%) | 1 (10.0%) | 10 | |
| Middle–late | Adolescent | 18 (52.9%) | 16 (47.1%) | 0 (0.0%) | 34 |
| Young | 41 (28.9%) | 83 (58.5%) | 18 (12.7%) | 142 | |
| Middle | 36 (23.2%) | 87 (56.1%) | 32 (20.6%) | 155 | |
| Old | 6 (13.0%) | 23 (50.0%) | 17 (37.0%) | 46 | |
| Unknown | 9 (25.7%) | 23 (65.7%) | 3 (8.6%) | 35 |
Abbreviations: TS0 = total score 0; TSL = total score lower; TSH = total score higher.
Frequencies of preauricular groove types during the early Edo period and the middle–late Edo period (right and left, in that order).
Frequencies of the preauricular groove classes during the early Edo period and the middle–late Edo period.
Change of frequencies of the preauricular groove types along the age groups, comparing the early Edo period and the middle–late Edo period (right and left, in that order).
Change of frequencies of the preauricular groove classes along the age groups, comparing the early Edo period and the middle–late Edo period.
We did not find any significant differences in the frequencies of preauricular groove types and classes between the kamekan and hayaoke groups. The results of the Fisher’s exact tests were not significant in the pairwise comparisons (Table 11, Table 12; Figure 9, Figure 10). We also found different trends between the two groups for the frequency of ‘no change’ type of both sides and TS0 in different estimated age groups. That is, in ‘adolescent,’ the frequency of ‘no change’ type and TS0 was higher in the hayaoke group than in the kamekan group, whereas in ‘old,’ it was higher in the kamekan group than in the hayaoke group (Figure 11, Figure 12). However, the results of the overall Fisher’s exact test of these frequencies showed that the observed differences were not statistically significant (P > 0.05).
Frequency of preauricular groove types (‘no change,’ ‘weak,’ and ‘strong’) in each of the two types of burial facilities categorized by age group
| No change | Weak | Strong | Total | |||
|---|---|---|---|---|---|---|
| Kamekan | R | Adolescent | 3 (33.3%) | 6 (66.7%) | 0 (0.0%) | 9 |
| Young | 20 (41.7%) | 24 (50.0%) | 4 (8.3%) | 48 | ||
| Middle | 13 (27.1%) | 28 (58.3%) | 7 (14.6%) | 48 | ||
| Old | 5 (25.0%) | 8 (40.0%) | 7 (35.0%) | 20 | ||
| Unknown | 6 (75.0%) | 1 (12.5%) | 1 (12.5%) | 8 | ||
| L | Adolescent | 4 (40.0%) | 6 (60.0%) | 0 (0.0%) | 10 | |
| Young | 20 (45.5%) | 20 (45.5%) | 4 (9.1%) | 44 | ||
| Middle | 10 (23.3%) | 25 (58.1%) | 8 (18.6%) | 43 | ||
| Old | 6 (28.6%) | 7 (33.3%) | 8 (38.1%) | 21 | ||
| Unknown | 3 (30.0%) | 7 (70.0%) | 0 (0.0%) | 10 | ||
| Hayaoke | R | Adolescent | 7 (58.3%) | 5 (41.7%) | 0 (0.0%) | 12 |
| Young | 16 (40.0%) | 17 (42.5%) | 7 (17.5%) | 40 | ||
| Middle | 17 (34.7%) | 25 (51.0%) | 7 (14.3%) | 49 | ||
| Old | 1 (7.7%) | 7 (53.8%) | 5 (38.5%) | 13 | ||
| Unknown | 3 (60.0%) | 2 (40.0%) | 0 (0.0%) | 5 | ||
| L | Adolescent | 7 (70.0%) | 3 (30.0%) | 0 (0.0%) | 10 | |
| Young | 16 (42.1%) | 21 (55.3%) | 1 (2.6%) | 38 | ||
| Middle | 13 (27.1%) | 28 (58.3%) | 7 (14.6%) | 48 | ||
| Old | 1 (7.1%) | 7 (50.0%) | 6 (42.9%) | 14 | ||
| Unknown | 1 (20.0%) | 4 (80.0%) | 0 (0.0%) | 5 |
Frequency of observations of the different preauricular groove classes in each of the two burial facilities categorized by age group
| TS0 | TSL | TSH | Total | ||
|---|---|---|---|---|---|
| Kamekan | Adolescent | 4 (40.0%) | 6 (60.0%) | 0 (0.0%) | 10 |
| Young | 18 (35.3%) | 27 (52.9%) | 6 (11.8%) | 51 | |
| Middle | 13 (25.5%) | 27 (52.9%) | 11 (21.6%) | 51 | |
| Old | 5 (23.8%) | 7 (33.3%) | 9 (42.9%) | 21 | |
| Unknown | 4 (33.3%) | 7 (58.3%) | 1 (8.3%) | 12 | |
| Hayaoke | Adolescent | 7 (58.3%) | 5 (41.7%) | 0 (0.0%) | 12 |
| Young | 13 (31.0%) | 22 (52.4%) | 7 (16.7%) | 42 | |
| Middle | 13 (25.5%) | 26 (51.0%) | 12 (23.5%) | 51 | |
| Old | 1 (7.1%) | 7 (50.0%) | 6 (42.9%) | 14 | |
| Unknown | 3 (37.5%) | 5 (62.5%) | 0 (0.0%) | 8 |
Abbreviations: TS0 = total score 0; TSL = total score lower; TSH = total score higher.
Frequencies of the preauricular groove types in the kamekan and hayaoke groups (right and left, in that order).
Frequencies of the preauricular groove classes in the kamekan and hayaoke groups.
Change of frequencies of the preauricular groove types along the age groups, comparing the kamekan and hayaoke groups (right and left, in that order).
Change of frequencies of the preauricular groove classes along the age groups, comparing the kamekan and hayaoke groups.
Our findings revealed that the frequency of more developed preauricular grooves (the ‘strong’ type and the TSH class) increased with age, while their absence (the ‘no change’ type and TS0 class) conversely decreased with age. Because the number of pregnancies generally increased with age, this finding is consistent with the hypothesis that the development of the preauricular groove reflects women’s experiences of pregnancy and parturition during the Edo period, as also observed by Igarashi (1992b) in female skeletal remains of the Jomon period. The ‘strong’ type of preauricular groove did not appear in adolescent pelvises but subsequently increased exponentially in the middle and old-age groups, indicating that this type of preauricular groove resulted from more pregnancies or childbirths.
Changes in the prevalence of preauricular grooves between the early and later Edo periodAn increasing trend of the frequency of developed preauricular grooves was observed with the progression from the early to the middle–late Edo period. The lower frequency of occurrence of the developed groove during the early Edo period suggests that the average birth rate was lower in the 17th century than in the 18th century and later in Edo. This trend was not attributable to differences in age structure between the two samples, as the finding was observed for all age groups. A simple explanation could be that the number of fertile women increased after the early Edo period. This result is not only consistent with the hypothesis that shorter birth intervals increased the potential for frequent births in more recent years, as inferred from the studies of the duration of the weaning process according to the nitrogen isotope in the Edo period (Tsutaya et al., 2014, 2019, 2020), but could also be considered direct evidence that fertility actually increased.
Commencing from the late 16th century, when Ieyasu Tokugawa entered Edo and developed it as a central city, male laborers gathered there from all over Japan until the early 18th century, resulting in a very high proportion of males within the population (Koda, 1938; Kito, 2010). In light of the observed increase in fertility and the gradual shift towards equivalent sex ratio over time in the originally male-dominated population structure in Edo, as previously noted, it can be postulated that the average crude birth rate in and after the 18th century was likely higher than in the 17th century. Considering that even Edo was in a state of population stagnation in and after the 18th century (Kito, 1996), it is highly probable that the population increase that occurred in Edo during the 17th century was caused by an influx from rural areas rather than by an increasing number of births of the city’s female residents.
From another perspective, the prohibition order against child abandonment in Edo in 1690 (Tsukamoto, 1983) could also have influenced the population reproduction trend. This prohibition, coupled with an increase in fertility, may have contributed to an increase in the rate of natural increase in the middle-late Edo period compared to the early Edo period.
Frequency of preauricular groove types and classes in the kamekan and hayaoke groupsThere was no significant difference in the frequency of preauricular grooves, either type or class, between the kamekan and hayaoke groups. As mentioned above, the kamekan group comprised mostly of individuals belonging to the samurai class, whereas the hayaoke group comprised mostly of common people. Therefore, our findings suggest that birth rates within different social classes in Edo have been similar since the 18th century.
Murakoshi (1993) analyzed historical documents listing the years of birth, marriage, and death of the family members of the feudal vassals of the Uwajima domain (a part of present-day Ehime Prefecture) and found that their total number of live births since the late 18th century was 4.3, which was the same level of fertility as that of the farmers. Despite regional and occupational differences between farmers and townspeople, the results of the present analysis based on human remains are consistent with this demographic study in that the birth rates of samurai and commoners were at the same level.
At the same time, the result that the frequency of ‘no change’ type and TS0 in ‘adolescent’ was higher in the hayaoke group than in the kamekan group, while that in ‘old’ was lower in the hayaoke group than in the kamekan group, may reflect the difference in reproductive strategies between samurai and townspeople.
Based on the analysis of documents similar to contemporary family registration, it is known that farmer women had relatively long birth intervals, and the suppression of the number of complete births due to long-term lactation has been proposed as a reason for this (Hanley and Yamamura, 1977). In contrast, Sawayama (2017) proposed that high-ranked samurai families eliminated the lactation period and shortened the birth interval by hiring nannies to ensure legitimate children, and that they adopted birth control once they had the requisite number of children to maintain their family throughout the analyses of diaries of local samurai families.
It has also been suggested that farmer women had a relatively late age at first birth due to their apprenticeship to the city (Hayami, 2001), whereas the feudal vassal women in the Uwajima domain had a younger age at first marriage than the average farmers (Kito, 1991; Murakoshi, 1993).
The differences in preauricular groove development observed in the present analysis suggest that there may be distinctions in the reproduction strategies between townspeople and samurai families similar to those observed between farmers and local samurai families. However, due to the lack of statistical significance and the unknown reasons for the higher incidence of ‘no change’ type and TS0 ‘old’ in kamekan, namely, the higher proportion of individuals who may have been nulliparous during their lifetime in ‘old’ samurai, we refrain from presenting a definitive interpretation in this paper.
In the future, we intend to further examine the differences in reproductive strategies by adding data of preauricular groove development in females excavated from rural areas as well as those assumed to have been from high-ranking samurai classes.
The following conclusions were obtained from this study.
First, the frequency of preauricular groove types among the estimated age groups indicated that the prevalence of pregnancy and parturition scars increased with age. Thus, the formation of a preauricular groove in individuals during the Edo period can be regarded as a positive indicator of conception and the experience of childbirth.
Second, a comparison of skeletal material from burials during the early Edo period with those during the middle–late Edo period revealed a higher frequency of developed preauricular grooves in the latter group in all age categories. This finding suggests that the average number of births was lower during the 17th century than it was during and after the 18th century in Edo. It also supports the theory that fertility in Edo during the 17th century was relatively low, and that Edo’s population growth in the 17th century was mainly attributable to the inflow from rural areas.
Third, there was no difference in the occurrence rate of preauricular groove types and classes between the kamekan and hayaoke groups. This finding suggests that the reproduction rate within Edo’s population exhibited no social or hierarchical differences. On the other hand, in ‘adolescent,’ the frequency of ‘no change’ type and TS0 was higher in the hayaoke group than in the kamekan group, whereas in ‘old,’ it was higher in the kamekan group than in the hayaoke group. These differences may indicate a variation in reproductive strategies between samurai and townspeople.
Using the methods developed by Igarashi et al. (2020), we clarified the frequency of occurrence of preauricular grooves during the early modern Edo period and presented possible interpretations of the results of the analysis. However, to explore the reproductive history and demography of past human populations further, it is necessary to consider wider factors, including the number of pregnancies and childbirths contributing to the formation of the preauricular groove.
One probable factor has been proposed by Waltenberger et al. (2021), who indicated that the narrow birth canal and posterior position of the acetabulum influence the formation of pregnancy and parturition scars, including the preauricular groove. Thus, the relationship between the formation of the preauricular groove and pelvic morphology needs to be investigated. In this context, Takamuku (2011) reported previously that the size and shape of the true pelvis has changed over time in Japan. Therefore, data on preauricular groove morphology should be re-examined from the perspective of pelvic geometry.
We are deeply grateful to Prof. Hiromichi Ando at the Faculty of Letters, Keio University, Dr Kazuhiro Sakaue at the Department of Anthropology of the National Museum of Nature and Science, and Drs Takashi Nara and Junmei Sawada at the Institute of Physical Anthropology in Niigata University of Health and Welfare for granting us access to valuable stored materials. We would like to express our gratitude to Dr Yuriko Igarashi of the Nihon University School of Dentistry at Matsudo, and Prof. Takao Sato of the Faculty of Letters, Keio University, for their insightful comments that helped improve this study. We would also like to thank Radhika Johari from Edanz (https://jp.edanz.com/ac) for editing a draft of this manuscript. This work was supported by a Japan Society for the Promotion of Science Grant-in-Aid for Scientific Research (grant no. 23KJ1906).
