Marriages are Made in Kitchens: The European Marriage Pattern and Life-Cycle Servanthood in Eighteenth-Century Amsterdam

ABSTRACT Due to methodological difficulties of historical research on women’s labor, little is known of women’s contribution to household incomes in preindustrial economies. This article is the first to use domestic servants’ wages, as documented in account books from the period 1752–1805, to estimate the capital that women could accumulate during their years of service before marriage. As such, it offers a new perspective on women’s contribution to household resources. Results show that servants working for the most well-off households in eighteenth-century Amsterdam could save a marriage budget that was between one-third and half of the capital that an unskilled man could save in the same amount of time. Furthermore, servants’ wages would in theory have been sufficient to support a family of four at the subsistence level, illustrating that women’s wages and potential savings cannot be ignored.


INTRODUCTION
In present times, the economic and social significance of women's earnings is increasingly recognized (World Health Organization [WHO] 2006). Although many qualitative works on women's labor history have appeared in the past three decades, quantitative analyses of women's remuneration for paid and unpaid work in preindustrial economies have remained scarce. This has above all been caused by the scarcity of sources and methodological difficulties (Sheilagh Ogilvie 2003; Jane Humphries and Jacob Weisdorf 2014) causing the historiography on earnings in the preindustrial world to be dominated by analyses of men's nominal and real wages. Consequently, scholars investigating related topics such as historical living standards principally rely on men's wages and are consequently in danger of painting a distorted picture of household welfare.
In recent years, scholars have started to take up the challenge of unraveling the history of women's earnings (Nicola Verdon 2002; Elise van Nederveen Meerkerk 2010; Jane Humphries and Carmen Sarasúa 2012;Humphries and Weisdorf 2014). This article aims to augment this strand of literature by exploring domestic servants' wages in eighteenth-century Amsterdam. This information will be used to estimate the capital that women could accumulate during their years of service before marriage. As such, this article additionally sheds more light on the meaning of "life-cycle servanthood" (Peter Laslett 1977) for women's contribution to household resources at marriage. According to ample literature, in Western Europe, serving was a standard phase in many girls' lives: it provided them with a marriage budget and a way to "escape the authority of their parents" (P. J. P. Goldberg 1986Goldberg , 1992; Tine de Moor and Jan Luiten Van Zanden 2010: 11-14). It has to be noted however, that life-cycle servanthood was not a worldwide phenomenon: it was rare outside Western Europe (although research on central Japan has shown similarities), and even within Western Europe there existed other types of servanthood (Michael Mitterauer 1990;Mary Louise Nagata 2005;Raffaella Sarti 2007. Life-cycle servanthood is an important component of the "European Marriage Pattern" (EMP) that was outlined by John Hajnal in the 1960s and that has been used by many scholars as a theoretical framework since then. Hajnal stated that "a high age at marriage" and "a high proportion of people who never marry at all" were characteristic for Western European countries, roughly to the West of the imaginary line between Trieste and St. Petersburg (John Hajnal 1965: 101). Moreover, newlyweds did not move in with the groom's parents but started new households without being constrained by the wants and needs of living-in kin (Jack Goody 1996). Peter Laslett (1988: 153) has argued that this practice of "neo-locality" cost a substantial amount of money that could cause "nuclear-family hardship." This problem could be overcome because boys and girls left their parental homes at a young age (between 15 and 18) and married relatively late: on average, women married at the age of 25 and men at the age of 27.5 (Hajnal 1965;Jan de Vries 2008;de Moor and van Zanden 2010). This period of roughly eight to ten years was used to save money for the formation of a new household. For girls, servanthood was the most common way to acquire such premarital savings.
The concept of the EMP has not gone unchallenged. Recently, the idea that the EMP caused economic growth, as has been argued by de Moor and van Zanden (2010) among others, has been contested by Tracy 69 Dennison and Sheilagh Ogilvie (2014). First, they state that the EMP was a phenomenon that was not homogenous; that there existed notable differences between European countries in terms of age at marriage, rates of celibacy, and household complexity; and that these three components "were not invariably associated with one another" (Dennison and Ogilvie 2014: 686). Second, they argue that the exceptional economic prosperity of England and the Dutch Republic cannot be explained by this marriage pattern because these two countries did not meet all the requirements of a "typical EMP country." Instead, they emphasize the role of nonfamilial institutions, like guilds, and the available work.
Nevertheless, numerous scholars have successfully used the EMP as an analytical framework to investigate a wide range of topics, one of them being life-cycle servanthood (Sheila M. Cooper 2004;Deborah Simonton 2011;Jacob F. Field 2013). Many studies on European women working as domestic servants in the households of others have appeared (J. Jean Hecht 1956;Sarah C. Maza 1983;Marybeth Carlson 1993;Bridget Hill 1996;Ogilvie 2003;Simonton 2004). However, although many historians have contended that this period in the life cycle was crucial for premarital savings for many girls, so far, quantitative analyses of this phenomenon have remained scarce in the historiography.
The question of women's earnings links to the debate about historical living standards. Robert C. Allen introduced the method of establishing so-called "welfare ratios" of several occupational groups around the world (2001, 2003, 2009, 2013a, 2013b; Robert C. Allen, Jean-Pascal Bassino, Debin Ma, Christine Moll-Murata, and Jan Luiten van Zanden 2011). These ratios show to what extent "a man working full time could support a family at the 'bare bones' level of consumption" (Allen et al. 2011: 26). This method has been widely adopted by scholars (Leticia Arroyo Abad, Elwyn Davies, and Jan Luiten van Zanden 2012). Recently, the real wage method has been criticized because it does not incorporate women's earnings into the analysis (Jane Humphries 2013; Humphries and Weisdorf 2014).
Humphries and Sarasúa argue that this method of calculating welfare ratios is based on a male-breadwinner model that has not been the dominant type of labor division throughout history. Women's and children's contributions to household income must be taken into account as well (Humphries and Sarasúa 2012). Similar arguments are made for regions beyond Europe, for instance colonial Africa (Michiel de Haas 2014). Allen defends his method by pointing out that Humphries' own research on household labor division during the British Industrial Revolution has shown that men's incomes were by far the most important (Humphries 2010(Humphries , 2013Allen 2013a). However, this does not justify Allen's use of the male-breadwinner model in preindustrial times. In fact, a large strand of literature supports the idea that women in the early modern period were highly active in the labor market, for instance in the Dutch This article contributes to both these strands of literature. First, to illustrate the importance of incorporating women's wages into the literature on historical living standards, servants' welfare ratios will be calculated with the help of Allen's method. The results show that servants would have been able to stay above the poverty line, even when they would have to provide for a family of four. Although in practice domestic servants were unmarried women who did not (yet) have a family to take care of, indirectly servants' wages can have had huge implications on household welfare that are overlooked by Allen's method. For instance, women regularly used part of their saved capital to invest in the establishment of their own business (van den Heuvel 2007; Danielle van den Heuvel and Elise van Nederveen Meerkerk 2014). Second, this research takes a careful step toward estimating women's saved marriage budgets in relation to those of men. It will show that most servant women could save between one-third and half of the amount of money an unskilled man could save in the same period of time. This means that, at least in theory, a woman was responsible for a quarter to one-third of premarital savings.

SOURCES AND METHODOLOGY
Finding information on women's paid and unpaid work in preindustrial times requires creative archival research. Historians usually need to look for other types of sources to investigate women's wages than they would when investigating men's wages. The road to useful information may be less straightforward, but it is certainly accessible. Thus, by thinking outside the proverbial box, scholars can respond to Humphries' and Sarasúa's plea to "rescue the history of women's work from its marginal, 'off the record' status" (2012: 40). The strand of literature based on such creative research methods has been growing in the past years and will, hopefully, continue to grow. For instance, Luisa Muñoz Abeledo (2012) has recently consulted census takers' notebooks to investigate women's labor force participation in nineteenth-century Spain. This source tells us more about women's working activities than the actual censuses, since in the latter, women's work was usually under registered (among others: Hill [1993]; Schmidt and van Nederveen Meerkerk [2012]). Jane Whittle (2014) has demonstrated on the basis of probate inventories that widows in early modern England were active economic agents, performing mostly unpaid work such as farming. The present research is greatly inspired by the creative use of archival material in such works.
This article is based on an analysis of eleven account books from the period 1752-1805 in which heads of households recorded their earnings 71 and expenditures. The heads of households were the likely authors because regularly, in all account books, certain expenses were referred to as "money for my housewife" or "pocket money for myself." Moreover, the handwriting was very consistent, which implies that only one person was responsible for keeping track of the household's finances. The account books provided systematically registered information on the dates of arrival and departure, half-annual wages, and the specific positions of the servants they employed. The analysis resulted in sample of nearly 900 payments to in total 115 servant women and 68 servant men.
In the Dutch Republic domestic service was traditionally a female occupation. Relative to surrounding countries, male servants were less numerous. Although in England and France servant women still outnumbered servant men, the latter were mainly hired to improve the social status of the household. In the Dutch Republic, servants were hired just to cook and clean, not to be part of an entourage to impress society (Carlson 1994). Moreover, Dutch servant men's wages were remarkably low, which made this job unattractive; men had more jobs to choose from than women and working as a domestic servant was not the most profitable choice. In short, the demand for and supply of servant men was relatively low. Still, their wages will be incorporated into this article to investigate gender discrimination within the servant staff. For the estimation of men's premarital savings, unskilled laborers' wages have been consulted.
Solely the servants of wealthier families are included in this research because detailed account books from households in the lower strata of society are hard to find. In fact, almost half of all the domestic servants working in Amsterdam (12,000 in the year 1800) were employed by families with a relatively high annual income of between 4,000 and 30,000 guilders (W. F. H. Oldewelt 1945;Herman Diederiks 1982). Together, the households in this income category -just like the households under scrutiny in this article -constituted only 2 percent of the entire population of Amsterdam during the eighteenth century. My sample is therefore representative of almost half of the domestic servants in Amsterdam. It is likely that domestic servants working for less well-off employers received lower wages. The conclusions of this research therefore do not concern servants working in lower social classes. Moreover, the sources only give information on the years the account books cover. It could be that the servants continued to work for the same employer in the years after the account book stops. Finally, nonmonetary accumulated capital remains unseen. Servants were likely to collect items such as furniture and clothing, adding up to their monetary savings. In fact, remuneration in kind, such as lodging, food, and clothing, was a considerable part of a servant's total income (Carmen Sarasúa 2004).

NOMINAL WAGES
Eighteenth-century nominal wages around the world appear to have been remarkably stable. Allen et al. have illustrated this for the cities of London, Amsterdam, Leipzig, Milan, Beijing, and Tokyo, based on unskilled workers' daily wages in grams of silver. Naturally, the amount of these wages differed considerably: wages in London were almost four times as high as those in Beijing. Still, until the late eighteenth century, nominal wages in Amsterdam were almost on par with those in London ). The differences between these two Northern European cities would increase from the end of the eighteenth century onward, when London rapidly outpaced Amsterdam.
The economic success of the Dutch Republic had its roots in the sixteenth century, when the Dutch economy experienced a remarkable growth continuing well into the seventeenth century. This development went hand in hand with a rapidly growing urban population of the province of Holland, particularly the city of Amsterdam. While Amsterdam accommodated 30,000 people in the year 1585, the population had already grown to 120,000 in 1632 and 219,000 in 1680 (Erika Kuijpers 2005: 9). Although mortality rates in Dutch cities were high, the urban population grew thanks to the many foreign and domestic immigrants drawn to this prosperous region. Throughout the seventeenth century, approximately two-thirds of the Amsterdam population consisted of immigrants (Ad Knotter and Jan Luiten van Zanden 1987). Their segment diminished during the eighteenth century, when half of the population was not indigenous (Jan Lucassen 1983Lucassen , 2002. Many immigrant women from Scandinavia, Germany, and elsewhere hoped to find a job as a domestic servant (Lotte van de Pol 1996; Jelle van Lottum 2007). Several factors contributed to economic growth such as the rise of markets, high agricultural productivity, secure property rights, and advancing technology. These developments made it possible, according to Jan de Vries and Ad van der Woude (1995: 693), for the Dutch Republic to become "the first modern economy" of the world.
This economic expansion profoundly influenced wages. De Vries and van der Woude found that daily as well as annual wages increased considerably during the late sixteenth and early seventeenth centuries. From the 1660s onward, this growth slowed until the end of the eighteenth century. Thus, throughout the eighteenth century nominal wages hardly rose (Leo Noordegraaf 1980;Hubert Nusteling 1985;de Vries and van der Woude 1995). De Vries and van der Woude illustrate this development with construction workers' wages (Table 1). Although Jan Luiten van Zanden (2002) has expressed criticism of De Vries' and van der Woude's methods, he does state -after analyzing wages in the textile industry and shippingthat construction wages more or less represented the overall trend.  Table 1 shows the annual nominal wages of unskilled and skilled laborers. Originally, the data collected by de Vries and van der Woude are daily wages in stuivers (1/20th of a guilder). To facilitate comparison with the annual wages of domestic servants, these daily wages are converted into annual wages in guilders. The calculation is based on the assumption that a year consisted of 270 working days (de Vries and van der Woude 1995).
Much less is known about the wages of domestic servant women. In Amsterdam, servants were paid half-annual wages in May and in October, which were also the times when contracts ended, and new servants entered their service (J. W. Bosch 1931Bosch , 1932. There were notable differences among the salaries of the various types of servants. Therefore, I have divided them into different groups based on the level of their wage at the beginning of their service and their gender (Table 2). 1 The group of "lower-ranked female" positions includes labor maids, linen maids, and seamstresses, 2 who generally received lower wages. Kitchen maids, nursing maids, and dressing maids received higher wages and are classified in the group of "higher-ranked female" positions. The third group consists of the servant men. The servants without mention of a specific function are classified as "unspecified." Table 2 contains information on each of these groups: the HISCO-classification code, 3 the quantity, the average annual wage, and the average duration of tenure. Table 3 shows the number of servants in several wage groups. Table 3 shows that most women in lower-ranked service positions earned between 50 and 80 guilders annually. This number fluctuates between onefifth and one-third of the annual wage of an unskilled male laborer in Holland (Table 1). Higher-ranked servants were paid between 70 and 110 guilders annually -between one-third and half of the wage of an unskilled laborer. However, servants received food and shelter from their employers on top of their salaries. Thus, in real terms, their earnings were even higher. Later in this article, such payments in kind will be added to the servants' monetary wages. Previous research on servants' wages in Holland has shown comparable results. Marybeth Carlson found that "sixty guilders per year was the mean wage for a domestic servant in Hoorn, forty to seventy in Gouda, and forty to sixty in Leiden during the eighteenth century" (Carlson 1993: 110-11). For the city of Rotterdam, she found that annual wages for maids were 40 to 70 guilders, and for "cooks" 70 to 80 guilders in the period 1771-80 (Carlson 1993: 112). Hence, the wages of lower-ranked servants in Amsterdam (Table 3) correspond to the wages Carlson found for domestic servants in other cities in Holland. However, the wages of Amsterdam kitchen maids were relatively high.

74
Let us now look at the relationship between the wages of servant men and women. Surprisingly, not all servant men were paid significantly more than 75  their female colleagues, although there were notable differences between the various households and functions. Based on the average payments shown in Table 2 we can conclude that the coachmen were by far the best-paid servant men followed by personal servants. But the unexpected third place goes to the group of female cooks. Research on eighteenthcentury Leicestershire has shown a far larger gender wage gap (Penelope Lane 2004). For the year 1783, Lane's results showed a 0.5 ratio -"the proportion of female to male wages" (Lane 2004: 116) -for regular female and male house servants, and a 0.54 ratio for female and male cooks. Using the same line of reasoning for my data resulted in a 0.73 ratio for the labor maids/house servants and 0.85 ratio for the dressing maids/personal servants. Unfortunately, I found no information on male cooks, but the ratio from kitchen maids' wages to the average male servant's wage is as high as 1.03. These findings tell us that on average, there appears to have been relatively little gender wage discrimination between domestic servants (with the coachmen being the exception). Table 4 further elaborates on maidservants' positions by showing information on the household level. It demonstrates first, that on average the servant women outnumbered the servant men in all households. These findings confirm the idea that servant men were less common than servant women. Second, wage gaps on the household level were small. The exceptions were employers 1 and 7. In the first case, the large wage gap is caused by a very significant wage increase of one personal servant, and in the latter case by the high wage of a coachman. The general picture that emerges when studying the account books is that within one household, the coachmen formed the top of the servant hierarchy, followed by kitchen maids and male servants. The regular working maids were at the bottom.

CAREER OPPORTUNITIES AND TENURE
The type of position was the most significant determinant of wages. This is in line with Jacob Field's conclusion about servants in rural England, that "the key variable in determining domestic service wages was if an individual was a senior servant -whose wages tended to be considerably higher than other members of the household" (2013: 268). However, Field shows that only servants who had worked for the same household for a considerable amount of time could obtain such highly paid positions. In contrast, this was not necessarily the case in Amsterdam. The average tenure lasted around three years. Only 23 percent of the servants in the sample served in the same household for more than four years, and only 8 percent for more than seven years. These findings are consistent with the idea that serving bridged the gap between youth and marriage. Short services enabled women to save money in this relatively short period of time.
Thus, there is no direct relationship between the level of the wage and the length of service. This means that acquiring a certain skill level within the same household was not necessary to obtain a higher-ranked position. Perhaps girls with such higher-ranked positions had already served for other households where they had attained a certain skill level. Indeed, previous research has shown that servants moved around between households as a strategy for improving their working conditions. For instance, Simonton has stated that girls in early modern Western Europe "might go into service at a young age, working their way up a servant hierarchy, either in a single household, or moving from one household to another to improve their position" (2004: 373). My sources do not reveal what happened to the servants who left and whether they used this strategy too. However, seeing that the average tenure only lasted three years, it is likely that switching employer was a common tactic in Amsterdam as well.
The account books do reveal that it was possible for servants to improve their working conditions within the same household. Getting promoted to a higher-ranked position was rare, but not unheard of. One account book mentioned: "for Steijntie, for three months as a labor maid: fifteen guilders, for three months as a kitchen maid: twenty guilders" (GAA, Archief van de Familie Brants en Aanverwante Families, inv. 114). This girl was promoted halfway through her half-year contract. Salary increase was a more common way to improve working conditions. This happened to forty-five of the women and nineteen of the men in my sample (about one-third) from both the lower-and higher-ranked functions. The extent of the raises differed considerably between the servants and there is no general pattern to be found. There were extreme cases like Steijntie, who received 50 guilders annually at the beginning, and 100 guilders at the end of a six-year service. Generally, however, raises were more modest. These findings show that domestic servants were not necessarily stuck in one position or one salary scale. Making a career was even possible in a short period of time, which made this job appealing for young girls wishing to save for a marriage budget.

REAL WAGES
To understand the economic significance of these wages, they will now be analyzed in the context of living standards. To this end, Allen's method of calculating welfare ratios is applied (Allen 2001(Allen , 2009(Allen , 2013a(Allen , 2013b). Allen analyzed unskilled workers' wages around the world in grams of silver to make them globally comparable. Next, he composed consumption baskets with which he determined how much it cost to live on a subsistence level. In these baskets, Allen included the quantity of food a person needed per year, based on the required nutrients and national cuisine and climate, and the necessary nonfood products such as soap, linen, candles, and fuel (Allen 2013b: 12). The prices of these baskets were likewise converted into grams of silver. One basket represents the barebones consumption needs of an adult man. Based on this information, Allen calculated the welfare ratios of several occupational groups (Allen et al. 2011: 26).
Welfare ratios are measured by dividing annual wages in grams of silver by 3.15 times the price of one man's consumption basket in grams of silver. This calculation is based on the assumption that an average family was composed of a man, a woman, and two children who together needed to consume three (male) consumption baskets annually. Per basket, 5 percent of the costs is added to include rent into the calculation. In the case that the outcome is larger than 1, the family would have been able to survive. A welfare ratio lower than 1 means that the family lived in dire poverty and could not provide the necessary goods (Allen 2001). For Amsterdam, Allen found that the average welfare ratio of an unskilled laborer in the period 1737-99 was 3.75 in a range of 2.36 to 4.61, with a median of 3.77 (Allen et al. 2011: 27).
In recent years, this bare-bones consumption basket has been heavily criticized. Humphries (2013) has argued, on the basis of present-day calculations by the Food and Agricultural Organization of the United Nations, that the amount of calories in these baskets could not have been sufficient to feed a hard-working laborer. Previous research on heights and weights of nineteenth-century men and women had already confirmed this suspicion (Sara Horrell, David Meredith, and Deborah Oxley 2009). Moreover, Humphries has pointed out that nineteenth-century British working-class families on average contained seven to eight children (Robert Allen and Jacob Weisdorf 2011; Humphries 2013). Therefore, multiplying one male bare-bones consumption basket by three does not come even close to a realistic estimation of a working-class family's cost of living. Allen has (partly) accepted the argument that this basket does not contain sufficient calories and composed a revised bare-bones consumption basket (Allen 2013(Allen , 2015Humphries and Weisdorf 2014).
Nevertheless, Allen's method is applied to the servant wages to see whether Amsterdam servants would have been able to stay above the poverty line by means of their own wages, while taking into account the payment in kind they normally received. As such, the outcomes show to what extent servants' wages were theoretically sufficient to support a family of four, without any additional income from other household members. Although this situation is purely hypothetical, as most domestic servants were young, unmarried women, it is worthwhile applying this method because it illustrates that a calculation based on men's income alone, highly underestimates households' living standards.
I have used the average bare-bones consumer price index for Amsterdam for the period 1737-99. A basket cost around 198 grams of silver per year, per man . To calculate the welfare ratios, the annual servant wages are converted into grams of silver. 4 Throughout the seventeenth and eighteenth centuries, a guilder contained 9.61 grams of silver (Sophia du Plessis and Stan du Plessis 2012). The formula for calculating the welfare ratio of a household with a husband, a wife, and two children is as follows: (Annual wage × 9.61)/(3.15 × 198) = welfare ratio However, calculating the servants' welfare ratios in this way would be an underestimation. After all, domestic servants received payment in kind in the form of food and shelter (Theresa McBride 1974;Joyce Burnette 1997). I have therefore added the annual price of a one man's bare-bones consumer basket to their annual wage (Table 5). Humphries and Weisdorf also used this method of adding payment in kind to the monetary wage to estimate the wages of women living with their employers throughout history. They argue that "[i]n any case, the estimated daily costs of living simply capture the expenses that an average person would otherwise hold, had she not been living in" (Humphries and Weisdorf 2014: 12). Table 5 additionally shows the average wage of an unskilled man from Amsterdam in the period 1737-99 . As mentioned before, nominal wages remained remarkably stable throughout the eighteenth century. It is therefore possible to use the average wage over a long period of time. Table 5 shows that with these incomes, all servants working for these welloff households would have been able to support themselves, a spouse, and two children at bare-bones subsistence level. These outcomes are striking, for it shows that ignoring women's incomes can result in a distorted image 80 of household welfare. Although most women no longer earned a servant's wage throughout their married lives, the capital they could bring into their new homes thanks to this job was significant. Surely, an analysis of the wages of all servant women in Amsterdam would result in a lower average welfare ratio because the servants who worked in the lower strata of society, and probably received lower wages, are not incorporated in this research.
A more thorough analysis of these wages is necessary to determine women's financial situation upon marriage in relation to that of their husbands. In the following I will investigate how much money servant women were, in theory, able to save during their years of service. These savings will be compared to (hypothetical) unskilled men's savings.

SAVINGS
Peter Laslett has argued that the practice of neolocality in the EMP area was the basis of "nuclear-family hardship" (Laslett 1988: 153). In these regions, coping with misfortune was harder than in areas where extended families were the norm. Supposedly, the social safety net was more solid in nonnuclear households because there was more (adult) kin nearby to help. According to Laslett, serving in another household was a way to overcome nuclear-family hardship. Recent research by Annemarie Bouman and Tine de Moor (2013) supports this idea from the demand side as well. They state that commercially extending the household was a way of dealing with nuclear-family hardship: taking in lodgers helped to overcome financial difficulties while taking in servants supported the household physically. In this way, live-in non-kin replaced the help otherwise given by live-in kin. In this line of argumentation, servants not only supported themselves in overcoming nuclear hardship, but additionally helped their employers to overcome this problem.
More theories have been proposed to explain how people in the earlymodern period prepared themselves for possible hardship. For instance, Jaco Zuijderduijn and Tine de Moor (2013) have shown that for households in sixteenth-century Edam, saving cash was the most common way to avoid a possible future of misery. During the eighteenth century, saving was still common practice, especially for unmarried girls living apart from their parents. According to Hajnal (1982), in the EMP area, young girls accumulated savings during their years of service. Therefore, it is likely that the domestic servants in my sample tried to save as much money as they could.
I have estimated how much money servants were theoretically able to save at a maximum. Unfortunately, my sources do not show how much of their wages these women spent on consumer goods during their service. We already know that they did not have to pay for food and shelter. However, it 82 was customary that they bought their own clothing, which I incorporated in the analysis (Hilde Bras 2002). Throughout the eighteenth century, people spent approximately 15 percent of their expenditures on textiles (Van Zanden n.d.). Departing from this estimation, 15 percent of the servants' wages was subtracted.
Furthermore, one could argue that both servants and male laborers probably had to support family members as well. From my sources, I cannot tell whether this was the case. Although servants lived outside their parental home, and often at great distances as well, we do not know what kind of agreements had been made between the girl and her parents about remittances. Research by Thijs Lambrecht (2013) on remittances in England, France, and Belgium has shown that landownership was a crucial factor. He found that for landless parents "intergenerational solidarity" was more difficult to arrange. Lambrecht concludes that in Belgium and France, remittances from children to their parents were much higher than in England.
Based on existing literature on Dutch domestic servants, it seems that young women could save most of their wages for their marriage budget. Previous research by Bras (2002) on domestic servants in nineteenthcentury Zeeland has shown that young girls living with their parents were legally obliged to hand in their whole salary. However, this was not the case when girls served in another household far away from home. Bras suggests that the further away a girl migrated to work, the smaller the chance there was she would send money home to her family. She also found that many nineteenth-century servants opened their own savings account (J. Diederich 1951;Bras 2002). Serving was not only a way to overcome nuclear-family hardship, but also a way to be liberated from parental authority (De Moor and Van Zanden 2010).
Moreover, during the early-modern period, many servants in Amsterdam were immigrants from the eastern provinces or foreign regions such as Germany and Scandinavia (Sølvi Sogner 1993;Lotte van de Pol and Erika Kuijpers 2005). Research on Amsterdam marriage records has shown that from 1600 to 1800, "60 percent of the grooms and 44 percent of the brides were not born in the city, and 36 percent of the grooms and 21 percent of the brides were not even born in the Dutch Republic" (van de Pol and Kuijpers 2005: 46). Although I do not know the places of origin of the investigated servants, the chance that they lived far away from their parents is considerably high. The same questions arise with regard to men's savings. However, since the objective is to investigate the relationship between future spouses' savings, and therefore, it is possible to ignore this issue for both sexes in the analysis. Future research on this topic is nevertheless very welcome.
The calculation is as follows. First, the annual wage is multiplied by eight years because, as mentioned in the introduction, this was the average 83 period women had to bridge between youth and marriage. Subsequently, these numbers are multiplied by 9.61 to calculate their savings in grams of silver. Finally, 15 percent of the average capital is subtracted to incorporate their expenditures on clothing and shoes into the analysis. The formula is as follows: annual wage × 8 × 9.61 × 0.85 = estimated capital in grams of silver The results are displayed in Table 6.
To determine a man's hypothetical saved capital, the wages of unskilled laborers are used, based on Allen's data. Again, these estimations are based on the average annual wages of the period 1737-99 and are converted into grams of silver. Because the average unskilled laborer may not have received food and housing with his wage, the price of one man's annual respectability basket is subtracted. 5 This leaves us with the amount of silver a man would have been able to save per year. Finally, the annual capitals are multiplied by eight because this is the period on which the capitals of the servant women are based. Table 7 shows the results.
The amounts of silver displayed in Table 7 represent the capital that an unskilled laborer could save in eight years' time based on the assumption that he only had to take care of himself on a respectable level. I do not claim that these outcomes were the general rule and that every single man 84 and woman indeed saved such an amount of capital before getting married. Rather, these result are an indication of premarital savings. How can these figures inform the debates on the EMP and women's earnings relative to their husbands'? I have shown that in a period of eight years, a servant in a wealthy household could on average save f 537.86in a range from f 390.41 to f 669.33 -if she bought her own clothing and did not spend her wage on anything else. An unskilled man could save f 1,490.38 in the same amount of time if he only had to take care of himself at a "respectable" level. This means that upon marriage, a woman roughly owned between a quarter and half of what her husband's capital was worth, and between one-fifth and one-third of the capital the couple owned together. It has to be noted that only the five seamstresses accumulated a saving as low as a quarter of the one their (potential) husbands owned. The savings of all the other servants were higher than one-third. Hence, the financial balance upon marriage was not 50:50. However, with this amount of money, women could have made substantial contributions to the formation of a new household. The next step is to investigate the significance of these wages for living standards. As mentioned in the introduction, servants' wages do not inform us directly about household welfare because married women rarely worked as servants. There are however several ways in which premarital savings can indirectly contribute to estimating households' living standards. First, Dutch married women frequently ran their own business independently from their husbands (van Nederveen Meerkerk 2008). Recent research by van den Heuvel and van Nederveen Meerkerk on eighteenth-century Leiden has shown that numerous married women applied for a permit to open a coffee and tea outlet. Some 80 percent of the married women in their sample did so within five years after marriage. Therefore, starting a business was a life-cycle event coinciding with marriage (van den Heuvel and van Nederveen Meerkerk 2010, 2014). Moreover, running a business was easily combined with raising children. Indeed, van den Heuvel and van Nederveen Meerkerk have shown that the larger part of the married 85 women with a permit were mothers. To be able to start such a business, women needed seed capital. Women's savings, which could be obtained by working as a domestic servant before marriage, were thus crucial for their source of income after marriage.
Second, women's experience as domestic servants was not merely a preparation for being a proper housewife, but for many married women in the lower and middling classes it also meant an entry into market work such as washerwomen, seamstresses, or even hostesses in the various hospitals and orphanages found in Dutch cities (Ariadne Schmidt 2008). Retired servants could even continue working for their old employer as casual laborers (Humphries and Weisdorf 2014). Furthermore, serving for a wealthy household in the prosperous province of Holland may have helped girls climb the social ladder through the practice of hypergamy ("marrying up"). Servant girls could have been lucky enough to find a suitable middle-class marriage partner, possibly by using the social network of their employers. Moreover, immigrant servant girls returning to their home country were often considered desirable marriage partners. For instance, Norwegian girls who had served for Dutch households had higher social status in their homeland: they were called Hollaendsker (Dutch women) and even wore distinguishing clothing (Sogner 1993). Their reputation also enlarged their chances of finding a job back home because they were considered to be well trained in the Dutch households. In this sense, life-cycle servanthood can even be considered to be human capital investment.

CONCLUSION
This article extends our knowledge of women's contribution to household resources in the preindustrial Netherlands by calculating domestic servants' premarital savings. Based on my results, several conclusions can be drawn in the light of several strands of literature. First, wage gaps between servant men and women were surprisingly low. In the servant hierarchy, a kitchen maid was generally placed above a regular male house servant in terms of turnover. Second, in eighteenth-century Amsterdam, it is clear that domestic servants' real wages, calculated by Robert Allen's method of estimating living standards, in theory were sufficient to support a whole family. Their welfare ratios were well above 1, which means that they would have been able to support themselves, a spouse and two children. Although this situation is hypothetical, this conclusion has major implications. It shows the significance of women's earnings for household welfare. Third, life-cycle servanthood indeed seems to have been a lucrative way of saving for a marriage budget. I have shown that most domestic servants could, at a maximum, save between one-third and half of the amount of money an unskilled male laborer -their likely marriage partners -would have been 86 able to save. This means that, assuming that both the husband and the wife had saved for a "marriage budget," a woman owned between a quarter and one-third of the total available capital upon marriage.
Crucial to my line of argumentation is that saving for marriage by domestic service may have been a "life-cycle event," but it was surely one that resonated after marriage. The saved capital may have been used to invest in a private business that in due course would provide the family with (extra) income. Moreover, recent research has shown that there were numerous other ways in which women could earn money after they had married. All in all, considering that early modern Dutch women were also very active in several other segments of the labor market, a realistic idea about historical living standards can therefore not be provided without taking women's earnings into account.

Corinne Boter Wageningen University and Research Centre -Rural and Environmental History
Hollandsweg 1 NOTES found irregular payments to other women referred to as seamstresses who were clearly not living in. These women are not included in my sample. 3 Within the scope of this article, it suffices to say that the Historical International Standard Classification of Occupations (HISCO) is an initiative aimed at coding historical occupations around the world to make them internationally comparable. This enables researchers, among other things, to look up a description of a specific occupation in several countries and to see whether different titles can refer to the same occupation. 4 The analysis is based on the 887 payments and as such is not based on average starting wages. Thus, individual wage increases have also been taken into consideration. 5 Because an extensive analysis of the prices of the adjusted bare-bones consumption baskets has not yet been provided, I used the costs of a "respectability basket," which provided more calories and more luxurious products.