The structure of food taste in 21 st century Britain

This paper draws on the concepts and tools of Pierre Bourdieu to construct a comprehensive model of the contemporary British “food space.” It uses multiple correspondence analysis to unearth a space structured in two key dimensions revolving around the lean and the rich. A host of supplementary variables are available to examine the relationship, or homology, between food tastes and broader alimentary dispositions, including orientations toward shopping, ethics and cooking. Indicators of social position re-veal the structuring of the space by economic and cultural capital as well as gender, but also, updating and nuancing Bourdieu's own model for 1970s’ France, by age, region, ethnicity and religion. Finally, the paper examines the relationship between position in the food space and physical, mental and existential wellbeing, demonstrating that orientation toward the less healthy and the less rich, corresponding with few resources, is, in some cases, accompanied by not only hunger and deprivation but profound worry and misery.

and fruit, and the fatty/heavy/strong/cheap/nourishing (low capital), epitomized by pork, bread, and pot-au-feu.
The second axis, corresponding with capital composition, polarized the healthy/natural/exotic (cultural capital), for example, yoghurt and raw vegetables, and the rich/fatty/salty (economic capital). Systematically related to this, however, was the household division of labor. The richer a woman is in cultural capital, claimed Bourdieu, the less likely she is to devote time to cooking rather than employment and general childcare. For that reason, she tends to prefer quick and healthy options, whereas the working classes, with women occupying traditional domestic roles, are more likely to devote time to preparing cooked dishes. Indeed, the attitude of the working class toward food and the body was, for Bourdieu, one area of relative freedom or resistance against bourgeois cultural domination: food was to be enjoyed, often in company, without restriction of portion or behavior. This was crosscut by gender, however (Bourdieu, 1984, p. 192): whereas the men were inclined toward the hearty, the nourishing, and the strong (red meat, cheese, alcohol, etc.), as signs of virility, women were typically more inclined toward the light, the small and the fiddly (e.g., fish, fruit, vegetables) even if they prepare heartier meals for their families.
Notwithstanding those who once touted the decline of class as a source of dietary differentiation (Mennell, 1995), research beyond France, including in the UK, has tended to confirm class differences in food consumption, with those in higher classes plumping for healthier, lighter options as well as expensive foods while those with fewer resources consume cheap, efficient and substantial meals (e.g., Charles & Kerr, 1988;Tomlinson, 1998;Wills et al., 2011). Extended from this is the more recent literature on "food poverty" or "food insecurity"-the most shocking contemporary symbol of which in the UK is surely the foodbank-highlighting the struggles, deprivations, and physical and mental plight of those so poorly resourced that they often have to go without food, or consume the cheapest, nutritionally devoid options, such as ready meals (Cooper et al., 2014;Garthwaite, 2016;O'Connell et al., 2019;Purdam et al., 2016). This is a far cry from the cheery and resistant image of working-class abundance set out in Distinction, raising the question of whether Bourdieu's brush painted too broadly, overlooking poverty, or whether the working class in Western societies has become more internally differentiated with the deepening of neoliberalism since the 1970s and, more recently, of austerity following the economic crisis of the 2000s (cf. Atkinson, 2017). At the other end of the scale, meanwhile, the explosion of culinary options and avenues in the wake of globalization has seemingly had the effect of spurring a search among the more advantaged for not just variety-sampling different cuisines and "exotic" restaurants-but also authenticity, that is, food with distinctive historical, geographical and socially connective connotations (such as that produced and sold by independent traders, farm shops and markets), and sustainability, that is, food bearing fewer air miles and eco-friendly packaging, as part of a broader concern with the value or ethics of food rather than its nutritional content or price (see, e.g., Johnston and Baumann, 2014;Paddock, 2014Paddock, , 2016Warde et al., 2019Warde et al., , 2020. While the existing research on food consumption certainly serves up a feast of novel insights and updates, however, the bigger picture tends to get lost. Studies are of food poverty or middle-class "foodies," of restaurant preferences or food expenditure, and so on, without putting each explicitly and directly in relation to one another beyond assumptions or vague references to the "others" against which certain practices are inescapably defined. This is despite the fact that each facet-this or that disposition, this or that practice, this or that purchase-only derives its full social meaning from its position within the system of food practices as a whole, that is, its opposition to other dispositions and practices but also its relative distance or proximity vis-à-vis the full cloud of dispositions and practices. When it comes to analyzing the relationship with class, moreover, there is a tendency to think unidimensionally: in qualitative studies, the focus is typically on key contrasts between those high and low in resources, while quantitative studies tend to mobilize hierarchical measures of social position (such as the Registrar-General's occupational scheme) and/or techniques (such as regression) unsuited to the reconstruction of the geometric distances and directions of differentiation. This is important because there may be internal class differences obscured by linear schemas and methods-a difference, for example, within the "middle class," and even among "foodies," between those happily consuming meat, the sign of affluence par excellence (Fiddes, 1991), and those consciously abstaining from meat on ethical grounds (e.g., animal welfare or environmental impact) linked to a certain symbolic mastery.
The concept of the food space is expressly designed to remedy these partialities. It situates elements in the whole by articulating the objective relationships between food tastes and practices in toto, and their multidimensional correspondence with capital in terms of both distance and direction. It is, moreover, perfectly mappable with the right technical tools, chief amongst which is MCA, a statistical technique designed to array variance between variable categories geometrically along multiple axes of varying magnitude. To date, however, there has only been one effort to deploy MCA to construct a comprehensive model of the food space: Flemmen et al.'s (2018) anatomization of culinary tastes in Norway. Based on data on food expenditure, they unearthed a primary axis opposing those with wide or eclectic tastes and those with more restricted tastes, which corresponded with income and household size, and a secondary axis opposing the healthy and the unhealthy, which bore a relationship with class as they measured it. On the basis of a tendency toward traditional folk foods among the higher classes, they reasoned that a search for "authentic" food may characterize those richer in capital. Although not an unreasonable conjecture, Hegnes and Gustavsen (2019) subsequently chastised them for drawing inferences their data did not readily allow, accusing them of jumping from expenditure patterns to dispositions too eagerly-a leap exacerbated, they said, by dependence on the household level of measurement, which obscures internal differences. 1 More than that, however, Flemmen et al.'s (2018) analysis remained focused only on the place of capital/ class in the food space. This is a legitimate focus, for sure-an effort to unveil which foodstuffs act as symbols of class position and relative value-but it leaves unexplored the multi-determination of food choice and, therefore, the food space. 2 For here is the nub: while Bourdieu posited a close correspondence between the social space and a space of lifestyles (of which the space of foods was an element), the former being a space of positions defined by capital and the latter being a second-order space of "position takings" or practices, mediated by dispositions, the correspondence is-as Flemmen et al. (2019) have rightly stated in another context-always an empirical question. This is because other factors can feed into the relational configuration of lifestyles and food practices and compete or blend with class to shape dispositions. Assessing the symbols associated with class position is one thing, but assessing the full complement of factors feeding into the structure of the food space and their configuration is a different-and overlooked-object of inquiry.
Bourdieu recognized the specific effects of gender on gustatory practices, of course-some of which have been broadly confirmed in later quantitative analyses (e.g., Neuman et al., 2019). Yet even his general model, as some have pointed out (Bennett et al., 2009;Rancière, 2004), neglected the role of ethnicity in shaping practice and complicating the relationship between class and lifestyles. 3 Bourdieu himself had suggested that classes are as defined by their ethnic composition as much as ethnic groups are defined by their class composition, true enough, and it is easy to fit themes of nationalist cultural exclusion into his overarching framework. Still, ethnic background can have autonomous effects on practice counteracting capital possession, and perhaps this is nowhere more pronounced than in relation to food, where cultural traditions and associated religious taboos bear on what passes peoples lips (Kershen, 2002;Panayi, 2008;Ray, 2004). The complicating force of ethnicity and religion is only likely to have become more pronounced since Distinction was published, moreover, with increased migration, the cultural diversification of national populations, and the internal class differentiation of ethnic groups-all of which, some have argued, have shaped the genesis of a relatively autonomous ethnonational space existing in complex combination with the social space (Atkinson, 2020;Hage, 1998;Tabar et al., 2010). In the British context, of paramount importance are not only the large Jewish community, the product of historic waves of migration from various lands to escape persecution, but the growing population, in the context of post-colonial flows and settlements of people, of Muslims, Sikhs, Hindus, and Buddhists-all of which have proscriptions or recommendations on eating meats and sundry cultural variations on substitutes.
Age may well have distinct effects on food consumption, moreover-opposing carefree youth to older people aware of their mortality, for example-as might regional specificities and familial obligations or influences (see, e.g., DeVault, 1994), even if these are all still modulated and refracted by class position in one way or another.
Ultimately, dispositions honed in more than one structural space (family, ethnicity, class), and meeting a certain field of possibilities given by spatiotemporal location, may harmonize or contend to generate practice and feed into the food space. Lahire (2004), recognizing this point, sees it as a problem for the Bourdieusian perspective, but, in fact, it only underscores that second-order spaces-of food, lifestyles, or political attitudes-are the outcome of adjustment to several spaces and, with that, the individual's "social surface"-their combination of dispositions across fields-rather than their dispositions, or habitus, attuned to just one space (Bourdieu, 2000; see also Atkinson, 2016a). Precisely how that multiplicity plays out in the formation of the UK space of foods, however, is unknown. Only by constructing a model of the space of foods, its associated dispositions, and the correspondence with various markers of social position will throw light on the matter-a task pursued in the following analysis.

| DATA AND ME THOD
The data used derive from the 2018 wave of the Food and You Survey (FYS) commissioned by the UK Foods Standards Agency and delivered by NatCen. The survey is a cross-sectional sweep of adults aged 16 or over in England, Wales, and Northern Ireland-the food tastes of the Scottish, therefore, are unavailable for analysiswith a final sample size of 3,069, of which 2,582 have been retained for analysis. 4 Being commissioned by a governmental institution, the core focus of the FYS is to chart the food behavior and attitudes of the populace so as to inform policy measures-it is, in other words, an instrument of state designed and interpreted by players in the political and bureaucratic fields to serve certain interests (Bourdieu, 2014)-the most practical consequence of which is the dearth of useful variables tapping into class position. While there are appropriate indicators of gender, age, household structure, ethnicity, religion and region, the only available proxies for location in the social space are a four-category variable for annual household income and a three-category variable for education level separating those with a degree and those with no qualifications from the rest, the last of these being labeled as "other" but essentially covering secondary-level qualifications and vocational post-secondary credentials. As limited as they are, however, these variables do at least cover the most efficient indicators of the two prime forms of capital in capitalist societies. They can, moreover, be crossed to produce a composite indicator that may give slightly sharper insight into capital composition as well as volume, though some categories have had to be aggregated due to very small relative frequencies (Table 1).
If the variables for social position are rudimentary, the array of indicators of food consumption and orientations is sufficient for constructing a fairly detailed model of the space of foods and associated dispositions.
The core, active variables mobilized relate to consumption of 17 foodstuffs, covering various meats, smoked fish and cooked shellfish, milk/dairy, cooked eggs, raw fruits, frozen fruits, raw vegetables (including salad), cooked

Relative frequency
Inc 1-2 Ed 3 <£26,000 p/a Degree 5.7 Inc 1 Ed 2 <£10,399 p/a Other qualification 3.9 Inc 1  vegetables, pre-packaged sandwiches, and ready meals. In addition, there are questions related to a whole gamut of activities and attitudes revolving around "the practice of eating" (Warde, 2016): • eating out: which meals (breakfast, lunch, dinner, any), where (restaurants, hotels, cafes, pubs, food vans), and considerations when choosing a place to eat (healthiness of the food, price, dietary requirements).
• cooking/food orientations: whether respondents "enjoy cooking," are "not interested in food," have time to cook and like to "try new things." • ethical/environmental considerations: whether there is too much plastic packaging on food and whether respondents check where the food they buy is produced.
• degrees of necessity and scarcity: whether respondents skip meals or go hungry (and how often), can afford healthy meals and have lost weight due to going without.
The questions on eating out and shopping locations are binary yes/no variables, whereas the questions on cooking/food and ethics comprise five-point Likert scales running from strongly agree (++) to strongly disagree (−−) through the "neither" option (n). The questions on necessity/scarcity are composed of either two or three categories. Finally, there are also several questions asking people to rate their general wellbeing, that is, life satisfaction, life as worthwhile, and happiness. 5 The 17 active variables are subjected to specific MCA to produce a multidimensional topology of modalities approximating the relational structure of the food space. The additional variables are then projected into the model as supplementary points and subjected to statistical tests to discern the underlying space of dispositions and, in the case of the socio-demographic variables, the homology with the social space and other structuring features. These tests include a formula to determine the coordinate significance or "atypicality" on axes and rules for judging whether distances between points are notable (>0.4) or substantial (>1.0) (Hjellbrekke, 2019;Le Roux et al., 2020;Le Roux & Rouanet, 2004;Lebart et al., 2006).
The active variables are frequency-based, asking respondents how often they eat a named item: "at least once a day," "5-6 times per week," "3-4 times per week," and so on down to "less than once a month" and "never." There are, therefore, no assumptions about esthetic or symbolic meanings attached to the foodstuffs-these will be suggested instead by association with the supplementary categories. Some categories denoting extreme frequency or infrequency of consumption have been combined to avoid diminutive relative frequencies (<5 percent) and overpowered modalities, with some variables thus having fewer constituent categories than others, but the variables are otherwise left untouched (Table 2). In the interest of legibility, the category for most frequent consumption in each case is labeled as 1, with subsequent frequencies being labeled 2, 3, etc.

| THE S PACE OF FOODS AND A SSO CIATED D IS P OS ITI ON S
The analysis reveals a structure satisfactorily described in three dimensions, though only the first two of these will be considered further here. 6 The premier axis bears a modified inertia rate of 43 and, judging from the modality contributions, might be said to revolve around an opposition between the varied and the restricted-like Flemmen et al. (2018) found in Norway-or, perhaps, relative abundance and scarcity. Either way, variety or abundance, and their opposites are defined largely in relation to foods that are either lean/healthy or rich (Tables 3 and 4). At the positive pole, therefore, stand those who tend to consume fruit, vegetables, and shellfish-but also sausages and duck/goose-fairly if not very regularly yet eat ready meals and pre-packed sandwiches less often. At the negative pole are those who seldom consume fruit, vegetables, fish, shellfish, and duck/goose, but also eggs, dairy, packaged sandwiches, and a range of meats.
The second axis, meanwhile, yields a modified inertia rate of 23. Here the opposition would appear to be between the lean/healthy and the rich/fatty/processed. At the negative pole stand those who eat fresh fruit and vegetables (raw or cooked) frequently but generally avoid fattier or processed meats and meals; toward the opposite pole sit those more inclined toward consumption of fatty and/or rich meats-duck/goose, burgers, sausages, precooked meat-as well as ready meals and pre-packed sandwiches, but who tend to engage in only middling consumption of fruit and vegetables.
Examining the relational configuration of the full range of modalities within the bi-dimensional space deepens and discloses the interplay of the two oppositions ( Figure 1). In fact, the distribution of some variable modalities across space-the tendency of both frequent and infrequent consumption of certain items (ready meals, sandwiches, sausages) to be associated with the negative pole of the primary axis, for example-make interpretation of the plane as a whole, partitioned into quadrants for convenience, imperative. The bottom right quadrant, for instance, is clearly characterized by regular intake of fruit and vegetables, and some consumption of eggs and fish, as well as an aversion to ready meals and sandwiches. It might thus be summarized as the zone of the lean and healthy. The top right quadrant, on the other hand, is associated with slightly less frequent consumption of vegetables and fruit (except frozen fruit) and a greater propensity to consume ready meals, sausages, burgers, and sandwiches, but also duck/goose and shellfish. Although seemingly quite a varied diet, there is perhaps some leaning toward the rich and fatty in this sector.
The top left quadrant comprises modalities registering, above all, rare or total lack of consumption of fresh fruit and vegetables (and fish) and preference instead for ready meals, burgers, and sausages. Starkly opposed to the bottom right quadrant, it might well be characterized as the region of the unhealthy, fatty, and convenient. The bottom left sector, finally, consists of avoidance of both meat/animal products and convenience foods. What exactly people here do eat is less obvious, though the relationship with the second axis suggests it may be fruit and vegetables. This region might therefore be characterized as not only anti-convenience but (largely) non-meat-an interpretation confirmed by the positioning of vegetarians and vegans in the space (Figure 2). Projection of the indicators of food practices and attitudes into the space reveals objective associations illuminating broader dispositional complexes within which food choices are embedded ( Figure 3). It transpires, for example, that the lean/healthy quadrant is closely associated with explicit concerns for health and ethics when it comes to food choices-those who show concern for the healthiness of eatery fare, where items are produced and the packaging they come in are typically situated here. Those who find pleasure in cooking, feel they have plenty of time for it and take great interest in food ("foodies") are also associated with the region, as are those who favor independent traders, markets and farm produce, all of which might be associated with notions of "quality," "authenticity," "novelty" or, insofar as it favors small enterprises over large corporations, the "ethical" and "conscientious." Food choices might thus be contextualized by a disposition oriented toward the non-essential and-in the shape of situating food in long-term or broad conceptualizations about health, the body, and ethics-the abstract; toward, that is, form over function, or a "bigger picture" over immediate nourishment.
However, the closeness of the modalities for independent traders to the axis line indicates that taste for artisan produce-as with the disposition for eating in restaurants-is shared by many of those in the upper-right quadrant. The higher one goes in the "rich/fatty" quadrant, though, the stronger the proclivity for eating out or "on the go" becomes-whether breakfast or lunch, a takeaway or food van, a café, pub or a hotel, a mini-supermarket or a garage shop. The concern for ethics-specifically in the form of the locale of production-dwindles too, and instead, considerations of price become more prominent, even if they are associated with paying for home delivery from a supermarket. Some here may well take a measure of enjoyment from cooking, but, all in all, food preferences would appear to be woven into a modus vivendi in which dining out is more prominent, whether it be a case of "picking something up" while "out and about" for work or leisure (e.g., a sporting event) or enjoying a sit-down F I G U R E 1 The space of foods, plane of axes 1 and 2. Note: Modalities with above-average contributions to axis inertia are marked *. All other modalities bear coordinates significant on one or both axes (p < .05) meal in a convivial environment. The taste and the surrounding yet immediate experience of food consumption, it might thus be inferred, assume greater significance here than elsewhere.
The upper left quadrant, where fatty/convenient foods are located, is characterized by an ambivalent or negative attitude when it comes to cooking and food ethics. People here typically either feel they have no time for food preparation, as if to indicate an element of pressure or necessity pervading everyday life, or are unable to say either way. The latter tendency, which also applies to trying new things and interest in food/cooking, may be related to the fact that not cooking for oneself is associated with the quadrant, or it may simply signal the relative insignificance of cooking and eating for many of those in the region-they are practical, functional requirements that one simply "gets on with" without either pleasure or worry. As for the ethical ramifications of food choices, they appear to be distinctly unconcerned with local production and either unsure or mildly in agreement that plastic packaging on food is excessive. Given that strong agreement on undue plastic is the modal response in the sample by some margin (70 percent), however, this actually translates into less concern than most. Again, this may be a symptom of the "principle of pertinence," as Bourdieu put it, orienting these individuals toward the practical business of getting themselves and their families fed efficiently rather than the ethical or health-related connotations of meals, which presuppose mastery of and luxury to adhere to abstract discourses on food.
Finally, the bottom left quadrant contains those with even less interest in food and cooking and, especially, those averse to trying new things. Most notable, however, is the association with not eating out in almost all its varieties. That this is not solely reducible to economic necessity, however, is suggested by the correspondence with special dietary stipulations when choosing a place to eat out-which, given the passive status of vegetarians or vegans in the analysis, refers primarily to religious proscriptions (halal, kosher, etc.). The alimentary practice may be guided, in this instance, by perceptions of the possible and the impossible based not (just) on a sense of material limits, therefore, but on the overt ethical principles of an explicit creed. A certain austerity and asceticism, then, mix with a distaste for the novel, the last of these perhaps indicating a certain traditionalism.
F I G U R E 2 Vegetarians/vegans in the plane of axes 1 and 2. Vegetarians (whole or partial) and vegans are represented by black triangular markers

| S TRUC TURING FAC TOR S
Analysis of additional supplementary modalities exposes systematic relationships with a variety of social characteristics ( Figure 4). Regarding the positioning of capital indicators, basic though they are, it transpires that those currently unemployed are disposed toward the cheaper, efficient, unhealthy options (such as ready meals), no doubt out of necessity, while the variable for income crossed by education displays a clear relationship with the primary axis. Variety/abundance is associated with higher capital volume, and its inverse, scarcity, corresponds with lower capital volume. There is some differentiation of the schema along the second dimension too, and though only two modality coordinates are statistically significant on the axis these are in direct opposition: those with the highest incomes but lower educational qualifications are associated with the quadrant of the rich and pleasurable while those with tertiary education but lower incomes gather in the healthy/ethical sector. 7 This could be construed as evidence of capital composition having some specifying effect on gustatory dispositions and practices, at least beyond the lowest region of the social space. One consequence of this would be that, in direct contrast to Bourdieu's argument in Distinction, a greater weight of cultural capital in one's stocks goes hand-inhand with (perception of) free time for and enjoyment of cooking while lower credentials and income are more closely associated with time constraints and distaste for cooking.
The homology with geographical space may be related to the relationship with class: the concentration of affluent and highly educated individuals in (non-urban areas of) the South East and South West of England probably has something to do with the location of these regions on the right of the space, and the tendency for those in the Northern areas to have lower incomes and lower qualifications could explain the association of these locales with F I G U R E 3 Tastes and practices in the space of foods. Note: Only modalities bearing coordinates significant on one or both axes (p < .05) are displayed. HD = home delivery; n=neither unhealthier food. This is not to discount the possibility of relatively autonomous geographical effects, however, such as a dearth of certain affordances (e.g., farm shops, independent fishmongers) for those who would otherwise be disposed to exploit them, on account of lack of regional demand, though the latter is still likely to be tied, in turn, to the class composition of the resident population.
There is, however, a whole series of additional factors corresponding with the space. The second axis, for example, appears to be structured by gender, with women tending toward the lower region of the space, occupied by the lean/healthy and the austere/vegetarian, and men tending toward the upper zone, where food is either something to be enjoyed or something functional. Single people tend to gather in the quadrant of convenience and functionality, where ready meals and burgers are regularly consumed, whereas those in partnerships of one sort or another, with or without children at home, are (just) more likely to gather in the lean/healthy quadrant. Perhaps this signals the relative insignificance of food for those without partners or children since it is not (yet) woven into care for others (or, by extension, of self), that is, the desires and obligations of a familial field featuring spouses and offspring. However, while the locations of different household structures in the space may all be statistically significant vis-à-vis one or both axes, the distances between points, as measured by standardized deviations, tend not to be notable. Another principle of difference with which household structure is intimately entwined, however, does bear notable distances across the second axis: age.
Although related primarily to axis 2, the younger/older opposition is skewed by axis 1 as well. The result is, overall, a correspondence between older age (and retirement) and the lean and healthy; between middle age and the rich/pleasurable; and between relative youth and the taste for functionality/convenience. It would be a mistake, however, to draw inferences from this purely in terms of age-specific factors-such as relative health status or temporal horizons-since the effects of age are clearly differentiated by capital possession. When crossed with F I G U R E 4 Indicators of social position in the food space. Note: Only modalities bearing coordinates significant on one or both axes (p < .05) are displayed education level, for instance, age categories are broken up and scattered across the space ( Figure 5). Older people are distinguished along axis 1 according to their acquired cultural capital: those with degrees are the most closely associated with the healthy/ethical quadrant while those with no qualifications gather in the austere/ascetic quadrant, where eating out is rare. All of those with no qualifications, in fact, are quite close to one another at the negative pole of the primary axis, as if to signify that similarity of capital possession severely attenuates differences by age at the lower end of the social space. Also striking, however, is the dispersion of those aged under 35 according to their qualifications. While those with mid-tier credentials remain close to the original location of the whole age group (they constitute over half the category), those with degree qualifications (37 percent of the age category) sit within the healthy/ethical quadrant, close to the modalities for using independent traders and displaying ethical concern. In this instance, the dissimilarity of capital possession entwines with a polarization of food tastes within the age group on both axes. Still, the consistent positioning of those aged 65 or over toward the bottom of the space and of middle-aged people along the center of axis 2 suggests age and its associated properties do bear some specifying force in structuring food taste.
This is not all. The austere/ascetic/traditionalist quadrant, where consumption of meat and convenience foods is limited, eating out rare and those with special dietary requirements typically located, is closely associated with

| PROXIMIT Y TO NECE SS IT Y AND MIS ERY
If everything suggests that the left side of the food space corresponds with fewer resources, including as proxied by unemployment, then perhaps it comes as no surprise to see that most of the starkest indicators of necessity are to be found here ( Figure 6). While they are all certainly in the minority compared to the sample as a whole (Table 5), those who-for economic reasons-have skipped meals (and regularly), gone hungry, generally eaten less and lost weight as a result, and who worry their food will run out before their next pay packet, whose food has in fact not lasted from one pay cheque to the next and who explicitly state they cannot afford to buy healthy food all gather in this zone. There is some relationship with the secondary axis, placing most of the modalities within the top-left quadrant of the space where convenience is paramount and necessity rather than asceticism or tradition the driving force, but the association is always strongest with the primary axis and, by implication, the lower capital that goes with it.
The physical effects of necessity are not difficult to imagine, and, indeed, those reporting "bad" or only "fair" health are typically situated in the left-hand sector of the space. The mental effects, given the level of worry associated with the quadrant, are plain to see too. More than that, however-though there is nothing in the wording of the questions to enable direct causal inferences-those populating this region of the space are the most likely to suffer existential pains: to feel unhappy, unsatisfied with their life and, ultimately, that life is not worthwhile. These are rare responses, for sure, but also associated with this pole of the space, and the capital stocks accompanying F I G U R E 6 Indicators of necessity and misery in the space of foods. Note: Only modalities bearing coordinates significant on one or both axes (p < .05) are displayed it, is the more common response of giving one's life only a "medium" score-an act which, given its relative nature, still qualifies as a sense of being unhappier and less worth while than others.

| CON CLUS IONS
The structure of the food space in the UK, as approximated by the variables available in the FYS, bears echoes of the space presented by Bourdieu in Distinction. Defined in terms of frequency of consumption, there is a prime opposition between relative variety/abundance and restriction/scarcity and a secondary, cross-cutting polarization of the lean/healthy and the rich/fatty. Going beyond existing analyses of food expenditure, moreover, and TA B L E 5 Indicators of necessity and misery

Label Definition a %
Often worried Often worried food will run out before getting money to buy more 4.1

Sometimes worried
Sometimes worried food will run out before getting money to buy more

12.3
Often not last Food often doesn't last, with no money to buy more 2.1 Sometimes not last Food sometimes doesn't last, with no money to buy more 9.7 Often not afford Often couldn't afford to eat balanced meals (last 12 months) 2.6 Sometimes not afford Sometimes couldn't afford to eat balanced meals (last 12 months) 8.9 Skipped meal Skipped a meal or cut portion size to save money (last 12 months) 6.9 Skip monthly Skip meals/cut portions almost every month 2.0 Skip sometimes Skip meals/cut portions some months but not others 3.3

Eat less
Have eaten less than I felt I should because there was not enough money (last 12 months) 6.9 Hungry Gone hungry but didn't eat because not enough money (last 12 months)

4.4
Lost weight Lost weight because not enough money for food (last 12 months)

2.9
Go without Not eaten for a whole day because not enough money (last 12 months) drawing together disparate threads of scholarship, it transpired that food consumption is systematically related to a complex of practices and attitudes indicative of varying dispositions: the abundant/lean with concerns for ethics, authenticity, experimentation, and enjoyment; the abundant/rich with eating out or getting nourishment on the go; the scarce/rich with convenience and functionality; and the scarce/lean with tradition, ethical nonconsumption of meat items and avoidance of eating out .
Variety and restriction are clearly related to the volume of capital people possess, but there is also some evidence to suggest the effects of capital composition in differentiating the lean/healthy/ethical from the rich/ fatty/pleasurable. Before proclaiming the existence of an unambiguous homology of spaces, however, it bears repeating that this differentiation is relatively modest and operates in conjunction with differences by gender, household structure, age, and ethnicity/religion. Perhaps with more refined indicators of capital, the relationship would emerge more prominently. On the other hand, it could simply be that food consumption is irreducible to a single logic, not even one as encompassing as class relations qua multidimensional social space, and that, instead, a multitude of social forces-pressures and desires, possibilities and obligations-feed into the choice to eat this or that thing on this or that day, whether the concern to cater for loved ones, to approximate orthodox images of desirable bodies, to fulfill religious observations and so on. If the food space bears any homology with the social space it is not because the latter is the sole determinant of taste and practice. The food space is structured by multiple factors or fields whose relative force and combination are open to empirical investigation. In short, food choices are an element of an individual's social surface, their complex of dispositions forged across social contexts, even if their class habitus can be said, from the evidence seen here, to be a key part of that.
A disconcerting finding is the association of physical and mental suffering with a lack of access to the healthy and the rich. Although the direction of causality cannot always be assumed-unhappiness born of some particular struggle can lead to lack of interest in keeping healthy rather than the inability to stay healthy breeding unhappiness (pointing once again to the multiplicity of possible generative principles of food practice)-in some instances, such as worrying about having enough food, it can. Those expressing such misery, moreover, are certainly a minority-though who knows how many people mask their malaise and "put a brave face on" in the interests of self-presentation when responding to survey questions? Yet the fact that any individuals in an affluent society go hungry, cannot afford the food they are told by "experts" to buy, and feel their life is unsatisfactory or not worthwhile-or even just less satisfactory or worthwhile than the lives of others-is disturbing, and the fact that they tend to be those divested of the fundamental sources of misrecognition in capitalist societies is an indictment of the rules of the game as it is currently played. Perhaps even more damning in the long run, though, are the consequences of the current situation for the health of the planet: so long as some (many) are denied the luxury to adhere to or, via educational inequality, the masteries to decode discourses on sustainability, the typically capitalist notion of "consumer power" as a fix for the environmental destruction wrought by centuries of human activity will remain a hollow mantra destined to fail.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available under licence via the UK Data Archive (SN 8,574).

E N D N OTE S
1 Hegnes and Gustavsen also criticized Flemmen et al. for adopting MCA, which requires (what they considered) "arbitrary" categorization of continuous variables, rather than principal components analysis, suggesting that MCA would be better suited to categorical frequency-based data-precisely the kind of data used in this paper.