1The optimal timing of weaning is widely discussed in terms of the so-called “weanling’s dilemma” (Lutter, 1992): the balancing between the known protective effect of exclusive breastfeeding against infectious morbidity, and the insufficiency of breast milk alone to satisfy the infant’s energy and micronutrient requirements beyond a certain age (McDade, Worthman, 1998; Kramer, Kakuma, 2009). During this period, infants first lose the immunological protection provided by breast milk (Newman, 1995) and second, they come into direct contact with environmental pathogens through complementary food, especially in the case of archaeological populations (Katzenberg et al., 1996). For all these reasons, timing and circumstances of the weaning process have great potential to impact not only the growth and health of an infant or child immediately (Wilson et al., 2006; Kramer, Kakuma, 2009; Lamberti et al., 2011) but also its development, health and physical well-being in the long term (McDade, 2005; Demmelmair et al., 2006; Haines, Kintner, 2008; Palou, Pico, 2009). However, as essentially a bio-cultural phenomenon, infant feeding practices are highly variable, with the age of weaning adjusted to different environmental, cultural, and economic contexts (Fildes, 1995).

Breastfeeding and weaning patterns in written sources

2It is particularly difficult to know breastfeeding and weaning practices for Early Middle Ages. Information found in written sources or iconography of 6^th-12^th centuries AD is exceptional1 (Flandrin, 1986; Morel, Lett, 2006; Rouche, 2006). For Early Middle Ages hagiographical texts are more eloquent on this point. Indeed, they relate about miracles performed by the saints during their life (Vitae) or after their death (Miracula), like healing or resurrecting miracles. These documents sometimes mention sociocultural practices, but often in term of prohibitions, like the taboo of the relation between sexuality and breastfeeding (Rouche, 2006). Then, hagiographical sources transmit a Christian discourse that tends toward a family ideal by male ecclesiastical elite, which may be far from real practices (Réal, 2001).

3Nevertheless, some tendencies can be highlighted in this literature, where the weaning is described as the interruption of breastfeeding (Lett, 1997, p. 91). Moreover, mentions of exact ages are exceptional, but in a few examples it seems to occur around 2 and 3 years of age, according to the miracles of Gregory of Tours (Rouche, 2006, p. 392) or the Byzantine Vitae (Bourbou, Garvie-Lok, 2009, p. 73-74, see also Chryssi Bourbou’s chapter in this publication). In the Eight Books on Miracles2, the age of 3 years appears as a reference of the weaning age for Gregory of Tours (538/9-594) in two stories: one tells us that a child was breastfed at the age of 3 years when he got ill3 while later the second one tells us that one should be 3 years old because he was still breastfed4 (Laes, 2011; Perez, 2013, p. 172). The later miracles of 12^th-15^th centuries confirm these ages at weaning (Krötzl, 1989, p. 30; Lett, 1997, p. 91), appearing as a gradual process in medical and pedagogical sources in a context of promotion of the breastfeeding, which becomes a major iconographic theme (Morel, Lett, 2006).

Stable isotopes analyses in infant feeding studies

4In an ecosystem nitrogen and carbon stable isotope ratios circulate through plants and animals. At each successive stage of the food chain, consumer’s tissues become enriched in the heavier isotope, resulting in δ¹⁵N and δ¹³C values that are 3-5‰ and 1‰ heavier, respectively, at each step, a phenomenon termed the “trophic level effect” (Minagawa, Wada 1984; Schoeninger, DeNiro 1984; Ambrose, Norr 1993).

5At birth, the isotopic composition of infant’s tissues is similar to those of the mother (Fogel et al. 1989). During breastfeeding infants derive their nutrition from their mother’s milk, and subsequently they exhibit an elevation of both δ¹³C and δ¹⁵N above maternal values. The isotopic values of tissues formed during the weaning process drop continuously with decreasing breast milk intake (Fogel et al., 1989; Fuller et al., 2006). For carbon, it was observed by Fuller et al. (2006) that the δ¹³C values decline to maternal levels more rapidly than the δ¹⁵N values, so Fuller and coworkers suggest they can track the introduction of the first supplementary foods, whereas δ¹⁵N values better reflect the length of breast milk consumption.

6Since the pioneering work by Fogel et al. (1989) stable isotope analysis is a routinely used tool in studies of weaning behavior of past populations. During the last years, the cross-sectional model, where the age of weaning is averaged across the whole population is often being replaced by a strategy of intra-individual sampling, which is to determine the exact feeding status of each infant (Eerkens et al., 2011; Waters-Rist et al., 2011). This enables us to avoid the source of isotopic variation occurring among adult females due to the differences in diet and/or physiology which, being replicated in breastfed infants, may limit the potential of stable isotope analyses. Additionally, applying this approach, we are able to describe the breastfeeding and weaning behavior in groups with low numbers of individuals (Herrscher, 2003; Kaupová et al., 2014). For these reasons, this approach was undertaken in case of Larina.

Material

7Larina (Isère, France) is an exceptional site about 30 km east from Lyon (fig. 1). It includes settlement and artisanal areas and two successive cemeteries from the 5^th to the 8^th century AD (Porte, 2011). The largest burial site of Larina-Le Mollard was used between the 6^th and the 8^th century AD. The graves are organized around a church of the middle of the 6^th century and the population appears to have been buried according to selected social criteria (fig. 2). Anthropological analysis of adults emphasizes changes in living conditions during the two centuries of the use of the cemetery. However, the archaeologists estimate that a one third of the graves were destroyed by ancient excavations from the end of the 19^th century (Porte, 2011, p. 314), but researches from 1977 to 1992 made the discovery of 447 graves, with a total of 402 individuals, 344 adults and 58 non-adults under 18 years (Porte, 2011, p. 450). A palaeopathological study was conducted especially on 57 subadults for whom an assessment of age-at-death was possible. The isotopic study was carried out on all the individuals whose state of preservation was sufficient to provide samples of bone and teeth tissues bringing the corpus to 12 individuals aged 0-5 years.

Methods

Osteological analysis

8Children represent a specific category of the population that has to be treated separately in the palaeopathological study, as illustrated in recent studies (Beňuš et al., 2010; Garcin et al., 2010). The study of non-adult skeletons has a certain advantage: some lesions occurring in childhood tend to remodel during growth and are no longer observable in the adult skeletons, leading to significant prospects for palaeopathological study of children (Scheuer, Black, 2000). However, osteological analyses can only be carried out on the excavated immature skeletons that are well-preserved and this is usually a minority of individuals. In Larina-le Mollard, using dental mineralization, an age-at-death was estimated for 57 subadults (Ubelaker, 1984), corresponding to 98 % of the excavated sample. To ensure the comparability of our results with previously published data (Herrscher, 2003; Kaupová et al., 2014), the 12 individuals included into the isotopic study (21 %) were re-aged using the Liversidge dental method (Liversidge et al., 1993, 1998). This method developed on the archaeological skeletal collection of Spitafields provides narrower intervals of dental age, which clarifies, but does not change, the picture of breastfeeding and weaning practices.

9However, age is not only a biological variable, but also a social fact related to behavioral changes bringing new risks, as illness, accidents or increased workload (Séguy, Buchet, 2013, p. 20). Indeed, medieval written sources reflect the changes of adult behavior towards children according to gender and age. Childhood is divided into stages reflecting a theorizing tradition of the lifecycle from Antiquity (Lett, 1997; Orme, 2003). For this study, the estimated biological age at death were distributed in social age groups using a division of childhood adapted to the Medieval society (Buchet, Séguy, 2008; Perez, 2013) (tab. 1).

Table 1 – Age distribution of the observable individuals of the studied sites

10The palaeopathological analysis focused on two non-specific stress indicators: cribra orbitalia and dental enamel hypoplasia. These anomalies of ossification and dental mineralization occurring during growth are the most commonly used markers to understand the health status of children (Lewis, 2007). Cribra orbitalia may have multiple causes and was first described as an indicator of nutritional stress, reflecting dietary iron deficiency, although this hypothesis has been revised. In 1992, P. Stuart-Macadam has shown that diet plays a minor role in iron deficiency and the cribra orbitalia should be considered as a response to environmental stress and a sign of adaptation to the environment (Stuart-Macadam, 1992). More recently, Walker et al. have shown that it may be related to infection or metabolic diseases, deficiencies of vitamin C or B12 or folic acid (Walker et al., 2009). In this study, cribra orbitalia was scored according to 3 stages on observable orbital plates: poor, medium and strong density (Perez, 2013, p. 287). Only 21 subjects with at least one preserved orbit have been taken into account, because of the poor bone preservation.

11Dental enamel hypoplasia is characterized by pits or horizontal lines on the tooth which may be linked to a number of factors, including fever, birth trauma, malnutrition, rickets, tuberculosis or deficient maternal diet, based on clinical studies (Lewis, 2007, p. 105). Dental enamel hypoplasia appears more frequently on canines, incisors and premolars (Goodman, Rose, 1990), but in the archaeological context, dentition is often incomplete. So, to avoid any representativeness biases, we counted as observable individuals those with at least 4 permanent teeth, as it is proposed in similar studies (Belcastro et al., 2007; Garcin et al., 2010). In these individuals, linear enamel hypoplasia (LEH) was noticed as present or absent on 32 individuals.

12If dental pathology is subject to large inter-individual variations, it provides a valuable source of information about life conditions and diet. In this study, dental pathology was observed in subadults on deciduous and permanent teeth. Dental caries and periodontitis were scored in 4 stages (Hardwick, 1960; Brothwell, 1972), dental calculus noted as present or absent. The rate of attrition of the teeth were determined in 5 stages (Brothwell, 1972). These lesions are only observed on erupted teeth and for the caries, individuals are considered as observable only if at least one molar is present (tab. 2). 427 teeth were investigated in Larina, 270 deciduous and 157 permanent, for a total number of observable subjects between 40 and 45 (tab. 2).

Table 2 – Results of the palaeopathological analysis

13We compared our results to 4 French medieval cemeteries located in Lower Normandy, which were investigated with the same methodology in a recent doctoral thesis (Perez, 2013): the rural cemetery of Saint-Martin-de-Fontenay (6^th-7^th centuries phases, Pilet et al., 1994) and 3 urban cemeteries from the 7^th to the 12^th century, as Michelet in Lisieux (7^th-8^th centuries phases), Notre-Dame of Rouen (11^th-12^th centuries phases; Niel, 2009) and Notre-Dame of Cherbourg (7^th-9^th centuries phases; Alduc-Le Bagousse, Pilet-Lemière, 1986). The number of observable individuals is presented in table 1.

Stable isotope analysis

14In our study, samples were taken from the bone at the mandibular border and from the apical end of the developing dental root. Bone undergoes constant turnover, such that its isotopic composition reflects the average diet for some period before death. The length of this period is dependent on the bone turnover rate, and consequently is highly variable and affected by different factors such as sex, nutritional status, physical activity and foremost the age of an individual (Martin, Armelagos, 1985; Seibel, 2003; Hedges et al., 2007). On the other hand dentine does not remodel, so its isotopic composition retains information about diet at the time of tooth development throughout life (Balasse et al., 1999). Therefore, once breastfeeding has started, the δ¹⁵N of newly formed dental tissues reflects the full trophic level effect in a short time, while the bone δ¹⁵N increases gradually along with the growth of bone and replacement of prenatally developed tissue. During weaning, for the same reason, the dental δ¹⁵N value drops much more rapidly to adult values (Herrscher, 2003).

Table 3 – Collagen preservation data

15The comparison of the values between “most recent” (dentine) and “some period prior to death” (bone) samples makes possible to estimate the nutritional status of a specific child on the basis of the relative nitrogen isotopic differences between tooth and bone samples (∆¹⁵Nt-b). A positive ∆¹⁵Nt-b (higher δ¹⁵N in tooth) suggests that the particular child was breastfed up to its time of death, a negative value (higher δ¹⁵N in bone) indicates that weaning took place sometime before. At this point, it should be stressed that weaning is not an overnight change of dietary status, but a continuous process starting with the first introduction of supplementary food and ending with the complete cessation of breastfeeding (Wilson et al., 2006; Pearson et al., 2010). Using this sampling strategy, we are not able to distinguish these two exact moments of the individual’s life, so that our terms “breastfed” and “weaned” should be interpreted with caution rather as “child still consuming breast milk substantially” and “child whose milk consumption decreased substantially” (Herrscher 2003; Kaupová et al., 2014).

16At least 30 mg of tooth dentine was sampled from the tip of the developing dental root at stage R_¼ -Ap_½, after (Smith, 1991). The samples were taken from different tooth types, regarding the age-at-death. This dentine sample reflects the diet immediately before death (tab. 6) with regard to a certain indeterminable period that the metabolic nitrogen pool takes to equilibrate with the diet (Balasse et al., 2001). Where possible, the first or second deciduous molar was sampled. From the oldest individuals (N=2) first permanent molars were used. Additionally, 50 mg of mandibular bone of each individual was sampled.

17Collagen extraction proceeded according to the Longin (1971) method, modified by Bocherens (Bocherens, 1992). Elemental analyses were performed on a Europa Scientific EA elemental analyzer connected to a Europa Scientific 20-20 IRMS at Iso-Analytical Limited, Crewe (UK). The uncertainty of isotopic measurements calculated on different standard replicates (IA-R042, IA-R045, IA-R046 and IA-R05, IA-R06) was less than 0,1‰ for both nitrogen and carbon. A feeding status through ∆¹⁵Nt-b was considered as a threshold for significance if the value was more than 0,2‰. The children with a ∆¹⁵Nt-b value higher than 0,2‰ were regarded as “breastfed”, those with a ∆¹⁵Nt-b value lower than -0,2‰ as “weaned”.

Results and discussion

Osteological data

18The distribution by social age group presents a high number of subadults from 1 to 7 years (tab. 1). However, some biases impact the representativeness of the sample, as incomplete excavations and graves’ destructions (1/3 in Larina-Le Mollard). The poor bone preservation results in low frequencies of studied subjects in the palaeopathological analysis: from 37 % to 79 % of individuals were observable for different lesions. The choice of an age estimation method has important consequences on the age distribution of a burial population (Scheuer, Black, 2000, p. 14; Séguy, Buchet, 2013). Sociocultural factors play an important role in the organization of graves. Indeed, some medieval cemeteries are characterized by a strong division of the space, according to biological and social criteria, such as age, gender, social status or kinship, as has been demonstrated in Rouen and Cherbourg, where only children under 7 years are buried around the walls of the church (Perez, 2013; Perez, 2015). In other cases, the number of infants is smaller than expected, partly due to burial of infants and juvenile outside the cemetery (Crawford, 1999; Gilchrist, 2012). Finally, non-adult skeletons are victims of a selective process, which may impact their age composition and so the representativeness of the skeletal samples.

19In the case of Larina, the relative absence of 0-1 year old seems directly linked to poor bone preservation, highlighted by the presence of shallow graves on the site and destructions of the ancient excavations. The high number of children from 1 to 7 years is more common in medieval churchyards: it may be partially due to the organization of graves of the children under 7 years around the church walls, where the excavations were done (Perez, 2013, 2015). Also, children less than 7 years are more affected by high mortality, because they are more sensitive to the pressure of the environment (Lewis, 2007, p. 85). The other sites included in this study present a similar trend in the age distribution, except for the rural site of Saint-Martin-de-Fontenay, where a lot of children aged 8 years or more were buried (tab. 1). Finally, these results do not represent mortality but only the distribution in age groups of the studied sample. Nevertheless, we can highlight some tendencies in the health status of children.

20Larina subadults present the lowest frequencies for all the lesions observed. The cribra orbitalia5 frequency is 29 % of the sample and linear enamel hypoplasia6 was observed in 31 % of individuals. In other sites, frequencies range from 48 % to 76 % for cribra orbitalia and from 20 % to 48 % for linear enamel hypoplasia (tab. 2; fig. 3). Then, in Larina 5 out 6 cases of cribra orbitalia are in stage 1 which is considered by some authors as benign lesion, too mild to represent serious health problems (Bennike et al., 2005). On the other hand half of the cases from the other sites are in stage 2 or 3. A similar trend is observed for dental pathology, except for caries (fig. 4). Indeed, 23 % of the subadults present caries7 lesions belonging to all stages. However, caries were found only on deciduous teeth, 5 and even 6 caries have been observed on the same individuals. The comparative series present also caries belonging to all stages on deciduous as well as permanent teeth, with frequency values from 11 % to 37 %. These results are in the average of medieval sites (Garcin et al., 2010; Lopez et al., 2012). Finally, the other dental lesions are less represented in Larina subadult sample: only 5 % present dental calculus, 9 % dental wear8 essentially in stage 1 and none of them suffer from periodontitis, in contrast with the other sites, which present higher frequencies of these lesions (fig. 4).

21The study of non-specific stress associated with dental lesions may indicate the health status, the life conditions and the resistance to environmental stress of medieval children. Subadults from the urban cemetery of Notre-Dame de Rouen (11^th-12^th centuries) are the most affected whatever the lesions. All the Normandy sites are geographically homogeneous and so we can formulate the hypothesis of a contrasted life conditions. Indeed, children from urban sites, as Lisieux, Cherbourg and Rouen present a worse health status, with higher frequencies whatever the lesions observed. In this context, the good health status of Larina’s children could be correlated to the rural environment.

22Our results highlight the importance to study caries on deciduous teeth, as confirmed by other studies (Alvarez, Navia, 1989; Kerr et al., 1990; Garcin et al., 2010). In fact, caries on decidous teeth may correlate to multiple factors, as swallowing problems (Vernois et al., 1989, p. 52) but also to the consumption of soft food composed of cereals (with carbohydrate), as soups or gruel. Then, a correlation may exist between malnutrition, dental eruption and caries since malnutrition during tooth development should have a greatest impact on the deciduous teeth (Alvarez, Navia, 1989). A child suffering from malnutrition since his first years of life should have less immunity and should be more sensitive to environmental stress and caries attacks. Medical studies on children from developing countries corroborate this hypothesis (Alvarez, Navia, 1989). In addition, more recent investigations underline other factors, as an addition of sugar in the bottle, of course not valid for our studies, or a prolonged breastfeeding (Kanchanakamol et al., 1996; Davies, 1998). In an analysis of 2200 children under 5 years, Mecky Matee found a correlation between deciduous caries, linear enamel hypoplasia and the breastfeeding practices: “the habit of allowing infants to sleep with the breast nipple in their mouths and the presence of linear hypoplasia were strongly associated with rampant caries” (Matee et al., 1994, p. 289). The presence of both lesions on the same individuals in Larina fits with this assumption, suggesting a stress during childhood. So, malnutrition, prolonged breastfeeding or specific habits could explain the importance of the deciduous caries in Larina sample.

Isotopic data

23All the samples met the criteria for good collagen preservation (tab. 3, fig. 5) (DeNiro, 1985; Van Klinken, 1999). Their elemental compositions range from 30,8 to 43,4 wt % for carbon and from 10,2 to 15,7 wt % for nitrogen. The collagen yield was in all cases higher than 1 % and C/N ratios ranged from 3,1 to 3,5. Based on the criteria described earlier (∆¹⁵Nt-b > ±0,2), the feeding status was estimated for 9 of 12 cases (tab. 4).

24In spite of a small number of observations, it appears that in Larina sample, the children aged 2 years or less (using the median dental age) had a ∆¹⁵Nt-b value higher than 0,2‰, which suggests that they were breastfed substantially (fig. 6). During the 3^rd year of life a strong decrease in the ∆¹⁵Nt-b value is observed. The first children were weaned at this age, while some others were still consuming a substantial proportion of breast milk. Due to the low number of older individuals, aged 4 years or more, on the one hand and a high variability of isotopic data, on the other hand, it is impossible to say precisely, when the process of weaning was completed. With regards to our findings on health status of Larina children, these results suggest that the subadult population of Larina profited from all the advantages of prolonged breastfeeding. By contrast, applying our sampling strategy, we cannot determine the exact age of first introduction of supplementary food, which surely took place sometime before the substantial decrease in breast milk consumption, which is possible to track in our data. This supplementary food may have protected children from potential nutritional imbalances, which could threaten the health of those that were breastfed exclusively for a long time (Kaupová et al., 2014).

Table 4 – δ¹³C, δ¹⁵N and their intra-individual differences with suggested feeding status

25At least in one case (LAR 11; 4,4-4,8 years), the ∆¹⁵Nt-b value is too high to be compatible with the scenario of prolonged breastfeeding, because due to the substantial proportion of bone, which had been formed under this nutritional regime, one would expect the ∆¹⁵Nt-b value to be close to the level of significance. This “postweaning” increase in¹⁵N in dental tissue in some individuals is regularly observed in the studies of this type (Herrscher 2003; Eerkens et al. 2011; Beaumont et al., 2013; Herrscher 2013; Henderson et al., 2014; Kaupová et al., 2014) and was also noted at the population level (Schurr, 1997). The dietary, physiological as well as pathological conditions may be considered as underlying causes of this trend (for more details on this discussion of elevated δ¹⁵N see e.g. Reitsema, 2013; Kaupová et al., 2014). The carbon results reveal an interesting trend of increasing isotopic values in the group aged 2-3 years, i.e. at the age when weaning is supposed to take place at least in a part of a population. This trend is contradicting the generally accepted scenario. According to Fuller et al. (2006) a slight trophic level effect in δ¹³C values (1‰) is also observable in breastfed infants. Subsequently, during weaning, carbon isotope values show a more rapid decrease to the adult mean than δ¹⁵N values, which led Fuller et al. (2006) to declare δ¹³C to be a good indicator for detection of the first introduction of supplementary food into a diet. However, the absence of any evidence of a trophic level shift in δ¹³C attributable to lactation is quite common (e.g., Herrscher, 2003; Waters-Rist et al., 2011; Howcroft et al., 2014; Kaupová et al., 2014). Other studies were inconclusive with some individuals supporting and others contradicting this scenario, (Dupras, Tocheri, 2007; Eerkens et al., 2011; Howcroft et al., 2012; Haydock et al, . 2013). This is also the case for Larina sample.

26Theoretically, the use of isotopically distinct weaning foods (such as those based on C₄ plants or marine foods) could blur a small trophic level effect of lactation in δ¹³C and cause the isotopic pattern observed in Larina sample. However, it is impossible to tease out this issue more thoroughly without detailed research on the dietary sources exploited at the site. Until today, only archaeozoological study was undertaken in Larina that found no evidence for the use of marine resources. Regrettably neither the isotopic analysis of adult/post-weaning diet nor detailed plant remains examination was performed up to date to Larina population (Columeau, 2011, p. 473-476). In the comparable sample of medieval population of Grenoble, which is temporal as well as spatially close to Larina, the general diet appears to be terrestrial and C₃ based (Herrscher, 2001), which makes a potential use of C₄/marine based weaning foods in Larina less probable. However a much bigger study sample of both adults and children would be needed to explore this issue thoroughly. With this sample size, one cannot exclude the possibility that the observed pattern is due to the small sample size.

27The main limitation of our study lies in the fact that the exact time necessary for recording the new nutritional regime in the isotopic composition of human dental and bone tissues is not known (Herrscher, 2003). In addition, there is a lack of certainty regarding how long the metabolic nitrogen pool takes to equilibrate with the diet (Balasse, 2001), and the bone turnover rate in the area of the mandibular base and time required under the regime of breastfeeding to balance the δ¹⁵N value in bone to those in the tooth dentine. For these reasons, the feeding status is sometimes difficult to establish because of the small difference between the isotope signatures of dental and bone tissues (∆¹⁵Nt-b < 0.2‰). Without the possibility to identify the cause, it may indicate an intermediate stage during the weaning process, or the lack of nursing. This case is observed for 2 children at Larina site. The same reasons also make it impossible to determine with certainty the threshold for the biologically significant difference between the tooth and bone. However, it has to be stressed here, that observed trends remain the same even if the threshold for significance is doubled from 0,2‰ (which is the double of analytical error) to ∆¹⁵Nt-b > ± 0,4‰.

28Our results show that there was not only one norm for the duration of breastfeeding established in the population of Larina. The evolution of the feeding practices through 3 centuries of the use of cemetery, the impact of sex or the social status may have contributed to the observed variability in weaning behavior.

29Present data suggest that Larina’s children may have profited from the advantages of prolonged breastfeeding, the youngest weaned child was 2-3 years old, but taking into account the “weanling’s dilemma” (Lutter, 1992), the apparently good health of these children can’t be wholly explained by applied feeding strategies. Other factors, such as rural environment, may contribute to this phenomenon. As suggested by Matee et al. (1994), prolonged breastfeeding may contribute to the relatively high caries prevalence in Larina sample. In future, isotopic analyses of other sites, especially those from urban areas, have to be carried out to tease out more precisely this question.

30Recent development in the field of stable isotope analysis may enable researchers to confirm these results by applying the methodology of serial section of the crown of first permanent molars (Beaumont et al., 2013; Henderson et al., 2014). Undertaking this sampling strategy, we would be able to increase our sample size by investigating the age of weaning in older individuals as well. This study confirms that the comparison between isotopic and osteological data enhances approaches to the study of health status and diet behavior in the past.