Pipe smoking and oral health in males from The Netherlands during the 18th–19th century

SUMMARY: The impact of tobacco on health is undeniable. It is a significant risk factor for multiple conditions, including oral diseases. There has been little research on pipe smoking and oral disease in the past. We compare caries, calculus, tooth loss, periapical lesions, and periodontal disease between rural 13th–16th century pre-tobacco males from Klaaskinderkerke and 18th–19th century pipe-using males from Beemster (N = 64). Pipe-smokers were more affected by all pathologies. Pipe smoking was a strongly gendered habit and possibly more common in the countryside. This work demonstrates a need to consider the impact of new imported behaviours on health in the past.


INTRODUCTION
In the late 15th and early 16th century, Europeans arrived in the Americas which resulted in the emergence of multiple new products and practices in Europe. Of the most significant was tobacco and the habit of smoking. By the mid-17th century, multiple European countries were importing millions of lbs of tobacco (Goodman 1993, 57). The Netherlands, in particular, imported vast quantities of the plant mostly obtained from trade (Gately 2001, 84), especially with the English. During this period the Dutch also grew their own tobacco (Roessingh 1976;Slicher van Bath 1968), making it widely available and eventually an inexpensive product. While the Dutch used tobacco in several ways, including snuffing and chewing, by far the most popular method of consumption was in the clay tobacco pipe, for which the Dutch eventually became renowned makers (Stam 2019). By this time, pipe smoking was extremely widespread; it is frequently depicted by the Dutch Masters in paintings of daily life (for instance Jan van Steen, Rembrandt van Rijn), in poetry and in writings (Brongers 1964, 137-151). Tobacco, the pipe, and pipe smoking became an important part of Dutch culture. For example, on marriage, couples were given a wedding pipe which they kept for their remainder of their lives together (Brongers 1964, 47). Tobacco became fully entwined with identities, and especially masculinity. Now, vast numbers of clay pipes litter post-medieval archaeological sites across the country (Duco 2008).
Today, the impact of tobacco smoking on disease prevalence and severity is unequivocal; it acts as a significant risk factor for multiple socially significant and costly diseases (WHO 2019). A number of conditions of the oral cavity have also been shown to have a strong link with smoking cigarettes (Axelsson, Paulander, and Lindhe 1998;Huang and Shi 2019), pipes (Albandar et al. 2000), and through second hand smoke exposure (Tanaka et al. 2015). These conditions include dental caries and antemortem tooth loss (AMTL) (Heng, Badner, and Freeman 2006;Vellappally et al. 2007), periodontal disease (Bergstr€ om 2014;Pej ci c et al. 2007;Sreedevi, Ramesh, and Dwarakanath 2012), and the build-up of mineralised plaque (calculus) (Albandar et al. 2000;Bergstr€ om 2005;Martinez-Canut, Lorca, and Mag an 1995;Sreedevi, Ramesh, and Dwarakanath 2012). In addition, tobacco use likely alters the oral microbiome which can create conditions conducive to the development of oral disease (Bizzarro et al. 2013;Huang and Shi 2019;Wu et al. 2016).
While there has been limited research on the relationship between tobacco use and health in past populations, evidence from historical sources suggests that people noted the relationship between poor health and chronic pipe smoking as early as the end of the 16th century (Charlton 2005), although there is little explicit mention of its deleterious effect on the teeth. Fortunately, it is possible to identify the use of pipes in past populations because of the abrasive nature of the clay pipe stems which abrade a characteristic 'pipe notch' into the enamel of teeth (see Fig. 1). The presence of these notches indicate that the individual was both a chronic pipe user and tobacco consumer. Although, there has been little dedicated research on the topic and work usually remains anecdotal, two published studies have compared oral diseases between those with pipe notches and those without, both of which have focused on urban poor Irish communities; Geber and Murphy (2018) focus on 19th century people from Kilkenny workhouse, while Walker and Henderson (2010) assess calculus in an 18th and 19th century community of Irish Londoners. Together, these have suggested a relationship between pipe-use and increased rates of caries, periodontal disease (Geber and Murphy 2018) and subgingival calculus (Walker and Henderson 2010). Both analyses have been valuable in demonstrating that in these 18th and 19th century Irish communities, pipe smoking was more common in males than in females (Irish Londoners males 23.4%, females 6.9%; Kilkenny males 60.7%, females 28.6%). As such, it seems possible that pipe use was a strongly gendered practice and, in Irish groups at least, gendered differences in oral pathology might be expected.
Despite the evidence that pipe smoking can significantly influence dental health, there has been no discussion of how the pipe smoking culture that emerged in The Netherlands impacted disease patterns, prevalence, and ultimately the physical health of people the in the post medieval period. Diachronic research on 17 different Dutch populations has shown that there were significant increases in rates of caries, tooth loss and periapical lesions in the Netherlands from the 16th to the 19th century (Constandse-Westermann, Nieuwenkamp, and Pot 2005). While pipe notches are sometimes incidentally reported, few authors present the true numbers of individuals with pipe notches, split pathology data by sex, or make dedicated comparisons between non-pipe using and pipe-using populations.
This research aims to assess how the prevalence of oral pathologies differs between pre-tobacco and posttobacco males from the rural Netherlands in order to assess whether pipe use influenced oral health during this period, with the hypothesis that they will be more prevalent in the post-tobacco group. As oral diseases can be strongly influenced by biological sex (Lukacs and Largaespada 2006) and pipe smoking can be strongly associated with masculinity (Brongers 1964), we focus on assessing differences in caries, tooth loss, periodontal disease, periapical lesions and calculus between males over time. Where possible we contextualise our data with that published elsewhere for the Netherlands, and where published we consider differences in prevalence of pipe notches between the sexes. Overall, this research presents the first direct results about the use of tobacco pipes in the 18th and 19th century in the Netherlands and in a rural community while also providing evidence about the role of pipe smoking in changing oral disease patterns in the post-medieval period. Furthermore, it will assess the health impacts of a pipe-smoking culture in a different community to those previously studied, which can be used to build a more detailed picture of the varying tobacco use practices in the past and its role in the construction of identities.

MATERIALS
This research compares individuals from two rural sites, Klaaskinderkerke (pre-tobacco) dating to the 14th-16th century and Beemster dating to the 18th-19th century (post-tobacco) (see Table 1). Both collections are currently housed at the Laboratory for Human Osteoarchaeology and Funerary Archaeology at Leiden University. The preservation of the material at both sites is generally good to excellent, although the graves occasionally suffer from some intercutting.
Klaaskinderkerke was located in the south east of the Netherlands on one of the small islands in Zeeland (Fig. 2). It was primarily a rural village involved in agricultural farming (Schats 2016, 24). Storms in the 1570s resulted in the village being abandoned prior to tobacco becoming widely available to the general population. Individuals from Klaaskinderkerke thus provide an ideal sample for this research as it is very unlikely that there was any pipe or tobacco use at this time. Human remains from Klaaskinderkerke were excavated in 1959 from Saint Nicolas Church. Fifty-four in-situ individuals were recovered including 33 men, 17 women and 4 children, in which no pipe notches were identified.
Beemster Polder is a Municipality in the north of the Netherlands, approximately 25 km from Amsterdam. While agricultural, the Beemster community were primarily involved in animal farming and they were famous for their wool and dairy product production (De Jong et al. 1998;Falger, Beemsterboer-K€ ohne, and K€ olker 2012). While generally most people were of rural working class, analysis of archival records shows a range of employments; not all were poor with middle class being present (Aten et al. 2012). In 2011, excavations by the Laboratory for Human Osteoarchaeology and Funerary Archaeology of Leiden University around the church in the village of Middenbeemster revealed over 400 individuals dating from the 18th and 19th century. This cemetery served all the communities in the Beemster polder. Around 63% were adults (over 18 years) and there was a roughly even male to female ratio. Although not systematically recorded, it was known prior to this work that a high proportion of males had pipe notches in their teeth making them ideal for this research. Significant quantities of clay pipes were also recovered during excavation. The present day community also remembers a tobacco drying house in close proximity to the village.

COMPARATIVE DATA
While there is little systematic data on the prevalence of pipe use in archaeological material from the Netherlands or elsewhere, there is some data for a few sites which are valuable for contextualising our research. For these sites it is generally possible to get information on caries rates and sometimes AMTL per person or by tooth over time, and in pre-and post-tobacco phases at one site (Catharinakerk, Eindhoven). Unfortunately many other sites could not be included because the dental data were not split by sex and we did not know the sex ratios in the utilised sample. The sites selected for comparison are outlined in Table 1.

METHODS
Both collections have been previously analysed by the Laboratory for Human Osteoarchaeology and Funerary Archaeology at Leiden University. Only individuals over the age of 18 years were included in the study since it was important to know the sex of the individual and sex estimation methods are not reliable in children. Age and sex estimates were taken from pre-existing data (Schats 2016). Here sex estimation was carried out using the standards of Buikstra and Ubelaker (1994) and WEA (1980). Age was estimated using scores from dental attrition (Maat 2001), cranial suture closure (Meindl and Lovejoy 1985), sternal rib end morphology (Işcan, Loth, and Wright 1984), the pubic symphysis (Brookes and Suchey 1990) and the auricular surface (Buckberry and Chamberlain 2002). Individuals were then categorised into age classes as follows: young adult (17-25 years), young middle adult (26-35years), old middle adult (36-45 years), mature adult (46 yearsþ). Holding a clay pipe in the teeth abrades the dental enamel over time leaving characteristic notches (Fig. 2). To be included in this study, the individual needed to have at least one quarter of the anterior dentition in alternating positions between the upper and lower jaw to ensure that all possible positions for tooth notches were observed. Although this significantly limits the number of observable individuals, it is the most robust way to ensure that people were correctly classified as pipe users and non-users. An individual was identified as a pipe smoker when at least one clear pipe notch was observed (0 ¼ no notch, 1 ¼ 1 or more pipe notches).
The Ogden (2008) method for recording periodontal disease was used because it considers the morphology of the alveolar margin rather than the extent of root exposure from vertical or horizontal alveolar wall defects. For reporting periodontal disease in archaeological skeletal populations, measurement of the depth of root exposure (in mm) is problematic because individuals in archaeological populations often have extensive dental wear which results in compensatory tooth eruption that furthers root exposure. Periodontitis was scored at each tooth position by documenting the alveolar bone form at the buccal margin following the ordinal scale in Ogden (2008): 0 ¼ unable to score (alveolar margin damaged or missing), 1¼ alveolar margin meets tooth at an acute angle (no disease), 2 ¼ alveolar margin is blunt with a slightly raised rim (mild periodontitis), 3 ¼ alveolar margin is rounded and porous, with a trough of 2-4 mm depth between tooth and alveolus (moderate periodontitis), 4 ¼ alveolar margin is ragged and porous, with an irregular trough or funnel >5 mm depth between tooth and alveolus (severe periodontitis). An individual's periodontal score was derived from the highest score observed in the mouth at any given tooth position.
The Brothwell (1981) method was used to record dental calculus on the ordinal scale of 0 ¼ none, 1 ¼ slight, 2 ¼ moderate, 3 ¼ considerable. The presence or absence of calculus was reported for each individual, and supra-and sub-gingival calculus were recorded separately following the recommendations of Brickley and McKinley (2004). An individual's calculus score was derived from the highest score observed on any tooth.
The Hillson (2001) method was used to record dental carious lesions. The method grades the severity of dental caries using an ordinal scale, where 0 represents no caries and 8 represents a gross carious lesion involving the pulp chamber (Hillson 2001). This method scores each tooth individually. A coronal carious lesion is recorded as gross caries if the lesion involves multiple coronal sites (e.g. the buccal and occlusal portion of the crown). Coronal carious lesions and carious lesions affecting the root or cementum enamel junction (CEJ) portion of the tooth are recorded separately following the Hillson (2005) scheme although for the purpose of this analysis they were counted together. Antemortem tooth loss (AMTL) was recorded when there was clear evidence for socket remodelling.
When making inter-site comparisons, to minimise the effect of interobserver error, we use only broad categories of data. As such, we only compare caries and AMTL rates. For the inter-site comparison of caries rates, we recalculated Beemster rates to only include those caries that penetrate the enamel and dentine. This ensured the data are more comparable to the other studies which do not use Hillson (2001) since this would likely lead to an overestimation of caries rates.
As these data are categorical, for statistical testing we used Fisher's exact test with Yates' correction factor for comparisons of numbers of individuals affected between pre-and post-tobacco groups, and for assessing variation in disease severity, we compared observed and expected values using chisquared tests. As oral diseases are age progressive, it is desirable to compare individuals in aged-matched groups, but there were insufficient observations to allow for statistically meaningful comparisons for most age classes. However, there were enough individuals in the young-middle age cohort at both sites to assess patterns in oral pathologies in an agedmatched sample helping to control for the impact of any age discrepancies between the samples. We did not statistically compare the young middle adults due to the low sample numbers. Statistical significance was set at p ¼ 0.05 and testing was carried out in SPSS 23.0.

RESULTS
In 2016, when our data was collected, 101 Middenbeemster men had some preserved dentition. Of these only 60 had anterior dentition that could be assessed for notches. In total we assessed 43 from Beemster where we could be confident of pipe use status and all 21 observable male dentitions from Klaaskinderkerke. Table 2 outlines the numbers of individuals in each age group. Due to our sample size, we assessed for differences in age between the two samples by placing individuals into two categories (younger adult (young and young middle (18-35 years)) and older adult (old middle and mature (36-45þ years)) using a Fishers' test, there was no difference in the age at death between the two sites (p ¼ 0.5884, n ¼ 64). However, it is worthy of note that there is a trend for Klaaskinderkerke males to be older at time of death.
Numbers and prevalence rates for oral pathologies are presented in Table 3. From Beemster we analysed 1023 teeth and 1256 tooth positions. For Klaaskinderkerke we assessed 311 teeth and 462 tooth positions. High levels of AMTL in Beemster males meant that a large number of males could not be included in this study because we could not identify whether they were using pipes or not. Surprisingly, all but one of the males from Beemster showed abrasion consistent with pipe use. While it is difficult to know the true prevalence of pipe notches in the population, due to the high rates of antemortem and post mortem tooth loss, it was notable that the minimum number of notches varied significantly. Some individuals had a single notch while most had multiple (see Fig. 3 for an example). As such, Beemster reflects a pipe-smoking population.
As can be observed from Fig. 4, the males from Beemster had a higher prevalence of all pathologies both on an individual level, and where tested, at the tooth level (see Table 3). These trends were also true for the young middle aged-matched cohort.
More Beemster males had caries than Klaaskinderkerke men, 97.7% and 61.9% respectively. The difference was highly statistically significant (p ¼ 0.0003, N ¼ 64). In fact nearly all Beemster males had at least one carious tooth. Furthermore, the number of teeth affected with caries was statistically higher in the Beemster group (47.4%) than in Klaaskinderkerke men (28.2%) (p < 0.0001, N ¼ 1355). A higher percentage of Beemster males had score 1-4 caries (55.8% v 23.8%) (p ¼ 0.0185, N ¼ 64), score 5 caries (48.4% v 19%) (p ¼ 0.0295, N ¼ 64), and score 6-8 caries (39.5% v 19%) (p ¼ 0.0156, N ¼ 64). Beemster males also had more teeth affected by score 1-4 caries (23.1% v 4.1%) and score 5 caries (6.5% v 3.2%) than Klaaskinderkerke men (score 1-4 p < 0.0001 N ¼ 1354, score 5 p ¼ 0.0262 N ¼ 1354). Gross carious lesions (score 6-8) affected slightly more teeth from Beemster (3.6%) than Klaaskinderkerke (2.2%) but the difference was not statistically significant (p ¼ 0.2800, N ¼ 1345). The trends were similar in  the aged-matched young middle cohort with the exception that the number of individuals with more severe caries (score 5 or above) was slightly lower than in the pooled-age sample, which is expected due to the age progressive nature of caries. The number of males with calculus was high in both groups, with all Beemster males and all but one Klaaskinderkerke man (90.5%) having at least slight calculus build up. As such, there was no difference in the proportion of individuals with calculus between the groups (p ¼ 0.1042, N ¼ 64). The same trend was observed in the young middle aged-matched cohort (Beemster 95%, Klaaskinderkerke 90%). The prevalence of slight dental calculus was higher at Klaaskinderkerke (supragingival 61.9%, subgingival 52.4%) than at Beemster (supragingival 34.9%, subgingival 32.6%). However, Beemster males tended to have more teeth with moderate and severe supra-and subgingival calculus deposits (see Table 3 for %s). The difference in the severity of supragingival calculus was significantly different between the groups (X 2 ¼ 6.441, 2df, p ¼ 0.039). We were unable to test the difference in subgingival calculus severity due to null observations in the considerable category for Klaaskinderkerke males. The same pattern was observed in the aged-matched young-middle age cohort.
More Beemster males had periodontal disease than Klaaskinderkerke males (Beemster 95.3%, Klaaskinderkerke 80%), however the difference was not quite statistically significant (p ¼ 0.0845, N ¼ 64). When considering the severity of the changes, similar numbers of men had score 2 and 3 periodontal disease (see Table 3), however more Beemster males had severe (score 4) periodontal disease, suggesting that the periodontal disease was more advanced at Beemster (Beemster 37.2% Klaaskinderkerke 23.8%). Statistically, there was no difference in stages of periodontal disease between the groups (X 2 ¼ 0.6950, 2df, p ¼ 0.7065). Again, like caries and calculus, the same trends could be observed in the young middle age-matched cohort.
Periapical lesions, which includes granulomas, abscesses and cysts, were statistically more common in the Beemster males, where 23.8% of individuals had at least one lesion (p ¼ 0.0068, N ¼ 64). This compares to 4.7% of Klaaskinderkerke which represents just one middle-aged man. This implies lower levels of dental pulp infection in the pre-tobacco males from Klaaskinderkerke. This result is supported by the findings in the young middle aged cohort where only Beemster males were affected.
The number of individuals with AMTL was statistically significantly different between the groups (p ¼ 0.0230, N ¼ 64), with 73% of Beemster males having lost at least one tooth in comparison to only 42.9% of Klaaskinderkerke men. Furthermore, the number of teeth lost was significantly higher (p ¼ 0.0006, N ¼ 1718) in the Beemster group (11%) than in the Klaaskinderkerke group (5.6%). This suggests that tooth loss was a much bigger problem for pipe-smoking males. This trend seems to start early as Beemster young middle adults are already more affected than those from Klaaskinderkerke.

COMPARISONS WITH OTHER SITES
Beemster and Klaaskinderkerke caries and AMTL rates are placed in context with other sites in broad chronological order in Table 4. Our data broadly fits with that of other contemporary sites with the exception that Beemster has very high pipe notch prevalence (97%). Generally, individual caries rate and caries rate per tooth seem to increase after 1300 CE. After 1500 CE, the number of males with caries seems to further increase, although the number of carious teeth remains stable or is lower. Although we have limited AMTL data, rates of AMTL also appear to increase in this period. Critically, in the posttobacco groups, the caries rates per tooth seem to show a negative correlation with the number of reported pipe notches. Females also appear to have higher caries rates per tooth than men in the post tobacco period.

DISCUSSION
The aim of this research was to assess whether the oral health of pipe-using males was worse than that of a sample of pre-tobacco males to evaluate the impact of the emergence of a pipe smoking culture on health. Before discussing the trends in pathology data it is worth commenting on pipe use. The high rates of pipe smoking observed in the Beemster males (here up to 97%) were unexpected. In broadly contemporaneous skeletons dating to 1650-1850 CE at Catharinakerk, Eindhoven, only 10 (9%) individuals had notches (Baetsen and Weterings-Korthorst 2013, 174). Higher rates were reported for urban Broerenkerk men (48%) from Zwolle, semi-urban Hattem males (north Netherlands) (60%) (Veselka and Klomp 2019), urban Roosendaal men (southwest Netherlands) (Veselka 2016) (59.1%), and in males buried in the Grote Kerk, Alkmaar (44%), although for the latter group it is suggested that many more may have been pipe users but tooth loss prevented scoring of notches (Baetsen 2001, 49). This was also a problem for Beemster. We observed an additional problem in young males. As the formation of pipe notches is age-related (Walker and Henderson 2010), at times we could not tell whether the wear seen in young individuals was or was not related to pipe use. This may lead to an underestimate of notches in some groups, but also an overestimate in our sample as we needed to exclude individuals when we could not be confident in pipe use. This problem can only be resolved with more systematic observation of notches using microscopy. Nevertheless, if the observed trends are a good reflection of the true incidence, then Beemster seems to be outstanding for its pipe use rates. Importantly, this is the first time that such data has been reported for a rural community, and it is possible that the habit was more common in farming communities than in towns. However, a further consideration that needs to be taken into account is the type of pipe likely to have been used by rural folk and the time period. Shorter pipes, known as cuttys, were preferred by manual workers as they could hold the pipe in the mouth and continue working with both hands, especially from the 18th century. Other types, especially pipes with longer, thinner, and more fragile stems, required at least one hand to be smoked. These may not leave such obvious marks on the teeth and might have been preferred in towns and were popular in earlier contexts. Further work on this is required and microscopy might help resolve this issue by assessing nonbiting surfaces such as the anterior (labial) surfaces of teeth which might come into contact with a pipe tip. Regardless, at all the Dutch sites where pipe use has been recorded, notches are rarely reported in women. While we did not record women at Beemster, they have only been noted in three cases during the routine assessment of the collection. This suggests that the habit was strongly connected to being male in the Netherlands in the 18th and 19th century. This is consistent with historic sources for the Netherlands that suggest it was not a habit common in women and it was related to masculinity (Brongers 1964).
All oral conditions were more prevalent at an individual level and at the level of the tooth in the pipe smoking males, and this was significant for caries, periapical lesions, and AMTL. Importantly, these patterns stood both for the pooled-age comparisons and the age-matched comparisons between the young-middle men showing that age differences between the two samples were not a factor in causing the differences. Furthermore, caries and supragingival calculus deposits also appear to be more advanced at Beemster. However, the difference in the severity of periodontal disease between the two populations was not statistically significant. Combined, there appears to be a relationship between disease and the use of pipes, inferring that the emergence of a pipe smoking culture was detrimental to oral health. This finding is comparable with the results of Walker and Henderson (2010) and Geber and Murphy (2018). This, and a number of other interesting results, warrant further interrogation.
Oral diseases are age progressive. In this study it was difficult to sample a sufficient number of individuals to perform a multiple age-matched comparison due to (1) the limited sample size of the Klaaskinderkerke sample, and (2) because many of the middle or older Beemster males had lost many, if not all of their of anterior teeth preventing the observation of pipe notches. When comparing the average age of our two samples, although not statistically significant, there was a tendency for the pre-tobacco males to be slightly older than the pipe-smoking males. Though initially this appears problematic for interpreting oral disease, it is in fact interesting; although the pre-tobacco sample skews older, these men actually have less oral pathology. This suggests that our data could underestimate the differences between the two groups.
When we contextualised our caries and AMTL data with that from other Dutch sites a number of trends emerge. Constandse-Westermann and Pot (1996) assessed changes in caries rate, AMTL and periapical lesions in 17 Dutch populations from 200 CE to 1850 CE. They found that there is an increase in all oral pathologies around 1200 CE-1300 CE, with a further increase from 1500 CE-1600 CE which coincided with the arrival of new habits and products in Europe including tobacco and sugar. The new data from our sites and the comparative collections assessed here generally fit Constandse-Westermann and Pots ' (1996) trends. The lowest rates of caries per individual were seen at rural Blokhuisen and Vronen from before 1300 CE. After 1300 CE but before widespread pipe use, the number of individuals with caries and caries tooth rate generally increases. After 1500 CE, the number of individuals with caries remains high, but caries rate per tooth stays similar or decreases, with the exception of the men in the post-tobacco phases at Catharinakerk, which have some of the highest rates recorded. While this might seem to counter arguments that pipe smoking does not increase caries rates, Beemster, Grote Kerk and Zwolle, where 44%-97% of males were pipe smokers, have much higher rates of AMTL  Ã This includes only caries that penetrate the enamel (e.g. Hillson 2001 scores 5-8).
þ Note that this is likely to be a significant underestimate due to the exclusion of males that have high rates of tooth loss that obscured the ability to observe pipe notches.
(Beemster 73%, 88% Grote Kerk, and 22.5% of all male teeth at Zwolle) in comparison to pre-tobacco groups and the post-tobacco phases at Catharinakerk. It is very likely that the caries rates at Beemster, Alkmaar and Grote Kerk are significant underestimates because of the high levels of tooth loss (see discussion below). Post-tobacco Catharinakerk is different with caries rates that remain similar over time with a small increase (13%) in AMTL. This finding is important because there is little evidence for pipe use at Catharinakerk with no cases in 1500-1650 CE and only 9% of individuals affected in the period 1650-1850 CE. As there is no reason to suspect that sugar use would vary significantly between Catharinakerk, Grote Kerk and Zwolle, as they are similar contemporary urban cemeteries, it suggests that pipe use is affecting tooth loss and caries development. This is supported by the work of Geber and Murphy (2018) who found higher caries rates for pipe users from within the same population. To help confirm these trends, it is important to produce methods that allow us to more precisely identify pipe use status of individuals so we can more confidently compare between using and non-using groups within the same populations to help control for other factors may be contributing to changes in oral pathology. More complete publication of oral pathologies by sex would also be informative and allow us to better unpick the impact of pipes versus dietary change.
Of the pathologies recorded, caries showed the greatest differences between our two groups. More Beemster men had caries, had more caries per tooth, and the caries were more advanced. Given that dental caries are a major entry point for bacteria that cause pulp infection (Hillson 2002), it is perhaps unsurprising that Beemster men also had more periapical lesions. Most clinical studies of cigarette smoking show a strong relationship between smoking and caries (Pej ci c et al. 2007;Wu, Li, and Huang 2019). Although the exact mechanism is unclear (see review by Vellappally et al. 2007), it is possible that the acidic nature of smoke may be detrimental to teeth. However, in our context this is unlikely to be the cause since past tobacco was more alkaline (Goodman 1993;Hughes 2003), as is pipe tobacco today (Pej ci c et al. 2007). Increasingly, research emphasises that other changes in the oral environment are likely responsible (Heng, Badner, and Freeman 2006;Liu et al. 2018;Wu, Li, and Huang 2019) including alterations to saliva quantity and qualities (Petru si c et al. 2015). Saliva plays an important role in removing debris, balancing pH levels and acting as an antimicrobial agent (Lenander-Lumikari and Loimaranta 2000). In particular, lower concentrations of salivary secretory immunoglobin A (sIgA) are associated with smoking and higher rates of dental caries (Hagh et al. 2013). Disruption in saliva composition and quantities could favour the persistence and multiplication of streptococci and lactobacilli, both of which secrete acids responsible for enamel demineralisation (Hillson 2002). Studies of oral bacteria from calculus in the compared populations here could enlighten whether our pipe-using individuals have excesses of bacteria linked to the cariogenic process. Regardless of the mechanism, as dental caries is a leading cause of tooth loss (Mays 2010), the high caries rates and their advanced nature may explain why Beemster males experienced more tooth loss. Importantly, this also suggests that the tooth caries rate in the Beemster individuals could actually be far higher than we are able to observe.
It seems likely that advanced caries may explain the high rates of tooth loss in pipe smokers, although it is possible that excessive attrition may also play a role. While we did not record dental wear in detail in our two groups, in an analysis of the data from Broerenkerk Zwolle, Constandse-Westermann, Nieuwenkamp, and Pot (2005) and Jackes (n.d.) recorded attrition by tooth type. Combined, they noted that males seem to progress faster through wear stages than women, especially in the anterior dentition, which they suggested to be caused by the abrasive nature of the clay pipes. To maintain bite, teeth will undergo compensatory tooth eruption. This, coupled with high rates of periodontal disease, could have caused greater loss of anterior teeth in pipe smokers. Abrasion of enamel from pipes could also be important for understanding the lower per tooth rates of caries they identified in the anterior dentition of men in comparison to women at Zwolle and it is hypothesised that this is due to the removal of demineralised enamel which effectively halts carious development. In fact, where data was available, post tobacco women did have higher caries rates per tooth than men. This may also partly explain the lower dental caries rates per tooth in Beemster in comparison to the other groups with lower pipe use prevalence, especially Catharinakerk, which has a very low pipe notch prevalence (9%) but the highest caries per tooth rate. It seems that within the tobacco using populations, that as male caries rate per tooth decreases, the percentage of individuals with pipe notches increases. This supports the hypothesis that pipe use did cause higher rates of tooth loss in pipeusing individuals but the mechanisms could be multifactorial and include caries and/or excessive wear.
While there was a non-significant trend for more Beemster men to have periodontal disease, the severity did not differ between the two groups. This result corresponded with that of Walker and Henderson (2010). This result was initially surprising since many studies in modern groups show that smokers have more severe periodontal disease including deeper probing depths, and greater amounts of attachment and bone loss (Albandar et al. 2000;Bergstr€ om 1999;Hahn et al. 1999;Sreedevi, Ramesh, and Dwarakanath 2012). It is possible that, given that we are dealing with pre-modern groups that had little in the form of dental hygiene, periodontal disease might have been so pervasive in Dutch populations before tobacco's arrival, that it is difficult to detect a change. However, recent clinical studies have shown that damage to the periodontal tissues may differ between smokers and non-smokers. Nicotine and other agents in tobacco smoke are vasoconstrictors potentially resulting in a loss of vessel density, a decrease in the number of cells involved in the inflammation process (Sreedevi, Ramesh, and Dwarakanath 2012), and reduced bleeding on probing (Albandar et al. 2000). This can lead to a reduction in inflammation, masking more advanced changes taking place in the oral cavity which can cause damage to soft tissue and bone (Sreedevi, Ramesh, and Dwarakanath 2012). It is possible that the method we used to record alveolar bone changes are not sensitive enough to distinguish between these two types of mechanism. Further observations of bone changes in skeletons of known smokers could help clarify this problem. An additional possibility for a lack of difference is the more alkali composition of pipe tobacco smoke, although Albandar et al. (2000) did identify higher rates of periodontal disease in modern pipe users. For now, it is not possible to say for certain whether periodontal disease was exacerbated by the introduction of pipe smoking.
Dental calculus was pervasive in both groups with nearly all individuals being affected. However, unlike periodontal disease, when we assessed the severity of calculus, Beemster males had greater calculus buildup at both the sub-and supragingival levels. This trend was also observed by Walker and Henderson (2010) in 18th/19th century London Irish smokers. These findings are also compatible with studies on modern smokers who generally have higher rates of sub-and supragingival calculus (Albandar et al. 2000;Bergstr€ om 1999Bergstr€ om , 2005Sreedevi, Ramesh, and Dwarakanath 2012) especially pipe smokers (Pej ci c et al. 2007). While it was initially thought that more severe periodontal disease would cause greater exposure of tooth roots to allow the build-up of subgingival calculus in smokers, it is now thought that the reduction of saliva and gingival crevicular fluid, an inflammatory exudate (Subbarao et al. 2019), may be responsible for subgingival calculus build up (Bergstr€ om 2005). This potentially provides greater opportunity for plaque accumulation and calcification through increased concentrations of calcium and phosphate in the saliva. This certainly fits with our data where we do not see significant evidence for worse periodontal disease in our smokers.
Overall, our findings are generally consistent with the two existing archaeological studies on pipe smoking and largely follow trends expected from clinical studies, even though the composition of tobacco smoke may have differed in the past. As we have been able to suggest that pipe smoking influenced oral health, it is possible that other important conditions of the period may have of also been exacerbated by the emergence of tobacco. Walker and Henderson (2010) already show that there was a relationship between pipe notches and evidence for respiratory disease. This demonstrates that when interpreting health in the post medieval period we do need to consider the influence of behavioural factors like pipe and tobacco use.
As tobacco smoking and pipe use was effectively an imported behaviour, its impact on health could also be seen as part of the increase in global connectivity and socio-economic priorities of the period. Initially, tobacco was touted as a disease panacea and was probably used to offer relief from a number of common ailments, including toothache (Vaughan 1612) which, based on our data, would have been common in our populations. However, it is also thought that very early on (17th century) smoking became popular with students at Leiden University (Brongers 1964, 19) and it spread throughout the classes as a social habit. It became an incredibly important part of sociability and was entwined with aspects of identity, especially masculinity in this context, but also intellect and worldliness (Hughes 2003). Tobacco proved to be highly lucrative generating significant revenues for business and state alike. So, while there is evidence for early opposition to excessive tobacco use in historical sources, sometimes based on health concerns (Charlton 2005) and/or religious (Burns 2007, 39), these concerns were not significant enough to prevent the development of the global tobacco market, something that has enormous implications for our health today.

CONCLUSIONS
This research aimed to assess whether emergence and development of a pipe smoking culture had a significant impact on the oral health of rural individuals in the Netherlands. Through the comparison of pretobacco and pipe-using samples it is possible to suggest pipe smoking led to a decline in health as all oral diseases, with the exception of periodontal disease, were statistically more common in pipe users. Contextualisation of results with data available from other sites, suggests that this decline was independent to those that may have taken place with the arrival of sugar, although further work with more populations and larger sample sizes are needed to closely control for dietary factors. In addition to this, rural individuals appeared to use pipes more and the practice was highly gendered with few females with pipe notches being reported for any sites. Future work should look to improve methods of tobacco use detection, either microscopically or through biomolecular approaches, which would help us securely identify who is using tobacco so that we can include individuals where it is not possible to observe the presence of pipe notches. This would be particularly important for understanding the role of sugar and tooth loss. As tobacco arrived as a direct result of the arrival of Europeans in the Americas, this work highlights how increasing globalisation and the formation of expansive trade networks can have negative effects on health.