Abstract
In the early-mid 20th century, sociologists were interested in sketching out possible similarities between caste and race. Since then, this approach has both been defended and – more commonly – rejected. In this paper, I contend that critics of the race-caste approach have fundamentally misunderstood the central arguments of that said approach, often committing methodological errors in their criticisms. I contrast this approach to advocates of the race-caste comparison. My argument is that this loosely knit group of scholars generated a sociological programme to think about racial domination as: (a) historically specific, which (b) in particular moments can appear to be caste-like, whereby (c) there are degrees of realisation of how rigid or loose the caste system can be (i.e. there is a caste continuum). Classical authors of the race-caste approach argued that the concept of ‘caste’ could help to generate comparative insights into hierarchical relations across time and space, and could provide insights into the mismatch(es) between attitudes, practices, and social structure in the era of burgeoning capitalist social relations. The central thesis of this paper, therefore, is that the race-caste scholars of the classical era developed a sociological programme oriented not towards demonstrating an equivalence between race and caste, but rather oriented towards the relational empirical questions of when, how, and why, in certain specific moments, racial domination becomes caste-like.
Comparing race and caste: ‘confusing and misleading’, or scientifically productive?
The idea that race relations in the United States could be compared to caste organisation in India is an approach that stretches back to the 19th century (e.g. Du Bois, 1898a). In the early-mid 20th century, sociologists in the United States (e.g. Warner and Davis, 1939) and India (e.g. Ghurye, 1969[1932]) sketched out similarities between the caste system as it functions in India, with especially the Jim Crow South in the United States (US). Since then, this approach has both been defended and – more commonly – rejected. From Cox’s (1945, p. 360) assertion that equating race and caste relations is ‘confusing and misleading’, through to Pal’s (2025, p. 55) declaration that comparing race and caste is a ‘terrible idea’, social scientists have been keen to reject the comparative approach to race and caste.
In this paper, I argue that critics of the race-caste approach have fundamentally misunderstood the central arguments of that said approach. I argue that critics often conduct multiple errors in their rebuttal of the race-caste approach, including their equating historically situated empirical descriptions with general definitions of caste, their flawed comparisons between ideal definitions of caste with actual empirical cases of race-relations, and their equating between the genesis, function, and legitimation of caste in specific contexts with a universal understanding of caste.
I contrast the critics’ approach to work of the classical social scientists who advocated for thinking about race relations as caste-like. I demonstrate how this loosely knit group of scholars generated a sociological programme to think about racial domination as: (a) historically specific, which (b) in particular moments can appear to be caste-like, whereby (c) there are degrees of realisation of how rigid or loose the caste system can be (i.e. there is a caste continuum). Advocates of the race-caste approach argued that their position allowed them to develop two main scientific gains. On the one hand, they argued their approach enabled them to generate a comparative sociology of hierarchy, that was able to compare: race relations across different local (e.g. the rural vs urban), national (e.g. US South vs North) and transnational (e.g. US vs Kenya) contexts, racial and non-racial hierarchical relations transnationally (e.g. race in the US South vs caste organisation in South Asia), and relations across time (e.g. pre and post-industrial periods in the US South). Secondly, they argued their approach enabled them to generate insights into the relations between beliefs, practices, and social structure – a key area of sociological inquiry. This was often realised in the way that scholars researched the mismatch between increasingly capitalist social relations versus the caste beliefs and practices which stemmed from a previous economic organisation.
The paper’s central thesis, therefore, is that the race-caste scholars of the classical era developed a sociological programme oriented not towards demonstrating an equivalence between race and caste, but rather oriented towards the relational empirical questions of when, how, and why, in certain specific moments, racial domination becomes caste-like. Furthermore, these scholars demonstrated that the answer as to when, how, and why racial domination can become caste-like cannot be answered a priori but must be ascertained empirically; scholars used the notion of caste precisely to show how it offers an empirical continuum to judge the rigidity of a given system of racial domination.
Caste and race: a definitional dilemma
To an extent, the debate about the comparability of race and caste is marred by definitional diversity – especially with regards to defining caste. Those who advocate for a potential comparison between race and caste often define caste in an almost Weberian ideal-type approach, providing a capacious definition of caste as a type of hierarchical organising of social relations. By contrast, those who vehemently argue against a race-caste comparison often do not define caste explicitly, but rather they use empirical description as a substitute for abstracted definition. Scholars on this side, therefore, often argue that caste is unique to South Asia precisely because they describe the caste system of that very region. From the very outset of this paper, therefore, it is clear that the ‘race-caste’ debate involves two parties who are working in quite distinct ways.
Consider, first, those who do advocate for a race-caste comparison. As I noted, such scholars tend to have Weberian-style ideal type definitions of caste as a type of hierarchical organising of social relations. While such scholars often noted that the ideal definition of caste was rarely fully realised in the empirical world, their definitions were deliberately structured such that we could get the most abstracted understanding of what caste is, and what caste does (Warner and Davis, 1939).
Indeed, Weber (1958) himself pursued this route when he viewed caste as a specific type of status group. Studying particularly Hinduism, for Weber (1968) castes are closed status groups organised into a rigid hierarchy, with each caste developing their own rituals and practices which distinguishes them from other status groups. Moreover, Weber (1958, 1968) argues, there are often ritualised, strict regulations on interactions between the higher and lower status caste groups – whether that be to do with physical contact, intermarriage, or separate religious and social spaces. Importantly, therefore, while Weber (1958) derived his understanding of caste from the study of Hinduism, by thinking of caste as a type of status group, he was able at least to abstract an understanding of caste such that it could be compared to other status-based hierarchical systems (for instance, he compared the relationship between caste groups and economic organisation in contemporary India to the feudal system of Mediaeval Europe).
Writing a couple of decades after Weber, Lloyd Warner and Allison Davis developed this abstracted definition of caste, again providing a capacious definition that would allow for caste to be brought into comparison with other types of hierarchical systems. Here, Warner and Davis (1939, p. 229) defined caste as ‘a rank order of superior-superordinate orders with inferior-subordinate orders which practice endogamy, prevent vertical mobility, and unequally distribute the desirable and undesirable social symbols’. This definition itself replicates Warner’s (1936, p. 234) earlier notion of caste relations as being characterised as a system of organisation whereby ‘privileges, duties, obligations, opportunities, etc., are unequally distributed’ across hierarchically ranked groups, ‘where marriage between two or more groups is not sanctioned and where there is no opportunity for members of the lower groups to rise into the upper groups or of the members of the upper to fall into the lower ones’. Writing again some decades later, Berreman (1960, p. 120) built upon Warner and Davis’s understanding to define the caste system as ‘a hierarchy of endogamous divisions in which membership is hereditary and permanent. Here hierarchy includes inequality both in status and in access to goods and services’. Much more recently, Loïc Wacquant (2024, p. 271) has synthesised the aforementioned literature to define a caste system as ‘constructed by (i) a rank-ordering of (ii) birth-ascribed, descent-based groupings that are (iii) endogamous, in which (iv) membership is permanent and transmissible, ands (v) whose hierarchical separation is justified by a collective belief in purity or congenital superiority’.
It is clear, therefore, that there is a tradition of social science which argues that ‘caste can be accurately defined in broader terms’ (Berreman, 1960: 120) such that it can be compared to alternate systems of domination (such as race) outside of South Asia. To such advocates of the race-caste approach, comparisons between race and caste relations are argued to be – in theory – possible, and at times, scientifically warranted and productive (Sharma, 1993).
By contrast, scholars who have denied the comparability of race and caste stress caste’s definitional specificity. However, the reasoning deployed in such critiques is often circular: caste is defined in such a way that it is rendered unique and incomparable, and thus the author concludes that caste’s unique nature means caste relations cannot be compared to race relations. Such circular reasoning is seen, for instance, in the work of Oliver Cox.
Indeed, Cox (1959, p. 3) begins his Caste, Class, and Race with the acceptance that there are ‘certain difficulties with respect to the formulation of a proper definition of caste’. The subject of what caste is, then dictates the opening two chapters of his book. However, nowhere in these chapters, nor in the rest of Cox’s corpus of scholarship (e.g. Cox, 1942, 1944a, 1944b, 1945, 1959), are we given an explicit definition of caste in the general. 1 Cox instead starts from the viewpoint that caste is unique to ‘Brahmanic-Indian’ society, and he then provides an empirical description of this said society, believing that a descriptive overview is a substitution for a definition of the concept. Thus, in Caste, Class, and Race, on the very first page, Cox (1959, p. 3) declares that ‘We are assuming, for the moment that a caste system exists in India only’, parallelling his earlier statement that ‘Brahmanic-Indian society represents the only developed caste system in the world’ (Cox, 1945, p. 360). Relying on historical descriptions of India, Cox (1945, p. 360) argues that caste is based in a centuries-long religious and cultural tradition, and is thus ‘ancient, provincial, culturally oriented, hierarchical in structure, status conscious, nonconflictive, nonpathological, occupationally limited, lacking in aspiration and progressiveness, hypergamous, endogamous, and static’. Thus, Cox renders caste unique and beyond comparison.
However, it is flawed reasoning to believe that historical elucidation of a concept (e.g. caste) in a specific case is the same as providing a general definition of the said concept. By comparison, consider if I were to try and define the concept of race, and I did so only by historically examining how race works in one case – for instance, colonial Haiti, where there were around 100 different racial categories (Getachew, 2016). If I were to do so, from this historical description, my definition of race would be too provincially tied to my descriptions of the Haitian context. This would lead me to make arguments such as that race is an organising principle for deciding how close one is to being a free/enslaved person – as per the logic of Haitian racial categories – which is not a principle of racialisation that remains true across all places and timescapes. In other words, given that Cox admits that he starts from the premise that caste is unique to India, and then defines caste through a description of caste in India, it should be apparent to us that his definition of caste may very well be provincially defined in virtue of his empirical scope. Indeed, this criticism of Cox is seen in Berreman’s (1960, p. 120) statement that ‘caste can be defined so that it is applicable only to India, just as it is possible to define narrowly almost any sociocultural phenomenon’.
Indeed, once we delve more into the literature which denies the incomparability of caste, we see that definitional issues are only the starting point. The literature denying the comparability of caste also has significant methodological issues in their argumentation, which I will review before proceeding to overview the work which defends the race-caste comparison.
Methodological errors in critics’ viewpoints
In particular, there are two main methodological issues made by critics of the race-caste approach. First, they often deny the race-caste comparison by comparing ideal definitions of caste with actual empirical cases of race-relations. Second, they tend to equate between the genesis, function, and legitimation of caste in specific contexts with a universal understanding of caste.
Caste in the ideal versus race relations in the concrete
With regards to the slippage between ideal definitions of caste versus empirical cases of race relations, this is seen in the historical just as much as the contemporary critiques of the race-caste approach. Consider Oliver Cox’s understanding of what he perceived to be immutable differences between race relations and caste systems. Cox’s comparison between caste and race often draws upon ideal-type understandings of caste in India with actual empirical (race) relations in the US, and in doing so, provides partial accounts of both cases. This criticism of Cox is similar to the problem Berreman (1960, p. 121) identified when he commented that ‘commonly, ideal behavior and attitudes in India have been contrasted with real behavior and attitudes in America - a fact which has led to a false impression of difference’.
Consider, for instance, Cox’s assertion that caste is stable in India, whereas race relations are often contested in the US. As Cox (1942, p. 223; emphasis added) argues: ‘Hinduism or the caste society of India is a powerful form of social organisation which may go on self-satisfiedly, so to speak, forever. It carries within itself no basic antagonisms. But the social aims and purposes of whites and Negroes in the South are irreconcilably opposed’.
2
For Cox, caste societies are perfectly assimilated, where there is no underlying contestation over social roles because the caste system is perfectly accepted by everyone. By contrast, he argues, race relations in the US are in a constant state of contestation, and even the dominant racial group (whites) often resort to abhorrent violence in order to maintain a racial status quo. As Cox (1945, p. 365; emphasis added) comments: The mob, nightriders, gangs, and ruling-class organizations such as the Ku Klux Klan; slugging, kidnapping, incendiarism, bombing, lynching, flogging, and black- mailing are among the principal reliances of the racially articulate whites for securing their position. The instability of this accommodation may be indicated by the fact that a goodly number of Southern whites are ashamed of it. In Brahmanic India, on the other hand, intercaste violence never takes such a pattern [. . .] The position of high-caste Hindus is guaranteed to them in Hindu society; but in America there is no such fundamental guaranty for whites. In the caste system there is social peace with the order; it is blessed in the most sacred books of the Hindus.
Here, we see Cox contrast his understanding of the caste system as it is supposed to function – where relations of hierarchy are perfectly naturalised and accepted – with the actual lived realities of race relations in the US. As Berreman (1960, p. 121) put it: ‘this contrast is invalid, resulting, as it does, from an idealised and unrealistic view of Indian caste, contrasted with a more realistic, pragmatic view of American race relations; Indian caste is viewed as it is supposed to work rather than as it does work; American race relations are seen as they do work rather than as they are supposed, by the privileged, to work’.
Furthermore, Cox is right to emphasise how race relations are constantly contested, but he is wrong to assume that caste relations are not. Scholars show that caste is often contested, whether that be ‘informally’ through privately disclosed criticisms of caste hierarchy (e.g. in Berreman’s (1960) research of lower caste anger and resentment in ‘Sirkanda’ towards the economic, political, and sexual abuse of power by the higher castes), or through public avenues – such as the legal system – as seen in the recent resistance of Dalit women to the economic practice of manual scavenging (Silliman Bhattacharjee, 2025).
Likewise, scholars such as Davis (1945) have argued that ethnoracial violence in the US tends to occur when whites perceive the caste order to be threatened. Davis (1945, p. 15) compares the old plantation areas in the South across Carolina, Alabama, and Mississippi, where ‘caste is almost “perfect” economically and socially’ to the newer industrial areas in mid-20th century Mississippi, where there was increasing economic competition between Black and white workers. Davis argues that white violence was more common in the areas where there was interracial competition and the supposed caste order was seen to be under threat, whereas in the areas with a lack of interracial competition, violence was not needed because the caste system was seen to be functioning well; as Davis concludes (1945, p. 15): In fine, where caste is most fully extended there is little need for violence, because the colored people are thoroughly subordinated economically, occupationally, and socially. When the castes are in economic competition as laborers and tenants, however, violence and conflict seem to be at their height.
We can thus use Davis’s work to respond to Cox’s (1942, 1945) assertion that racial contestation disproves its similarity to caste in India. Not only does Cox overlook the contestation of caste in India, but he also fails to see that in many cases in the US there is precisely an iteration of that ‘ideal’ type of societal equilibrium which he thinks is typical to the Indian context, where violence is not required because the dominant group have managed to maintain a rigid caste separation across all areas of social life. The white violence which Cox uses as evidence to show that racial relations are not caste relations can simultaneously be used as evidence – as Davis (1945) does – to build a hypothesis that racial violence tends to occur when the caste system seems under threat. As Davis (1945, p. 14; emphasis added) argued, ‘white society, as a whole, often resorts to terrorization to reassert the dogma of caste and to indicate that in physical and legal power over the life and limb of colored people, at least, the caste sanctions are effective’.
Related to the above critique, we can see that Cox often assumes that caste in India maintains this ‘ideal’ equilibrium because he regularly equates attitudes with beliefs; however, this is fallacious reasoning. Cox (1945) argues that how people act is directly representative of their belief patterns; the fact that, for instance, lower caste people in India show deference to higher caste people is then used as evidence by Cox that caste is socially accepted in India in a way that race relations are not in the US. As Cox (1945, p. 364) summarises: Negroes and whites in the United States stand toward each other in [. . .] a relationship implying suspended conflict. Conversely, castes stand toward each other in the relationship of superiority and inferiority, a relationship implying natural, socially accepted, peaceful status-ordering of the society. In the first case we have a power relationship in which definite aims and ends of each group are opposed; the second is a situation of mutual emulation or symbiosis among little status-bearing groups.
However, there is no empirical evidence to suggest that intercaste deference in India necessarily implies an acceptance of the caste hierarchy, and indeed in the US we often see interracial deference as a strategic means of avoiding interracial violence. In fact, this issue of intercaste deference was empirically investigated by Berreman (1960) in his research in India, in a place he names Sirkanda, situated in Uttar Pradesh. Here, Berreman (1960) found a caste system whereby the dominant Brahmans and Rajputs owned the majority of land, and a series of artisan castes would then work on this land with significant profits flowing to the land-owning dominant castes. In his fieldwork, Berreman (1960, p. 125) found that those of the lower caste – despite public displays of deference to members of the higher caste – harboured deep resentment towards their caste position, and had specific criticisms of the ‘economic, prestige, and sexual impositions by the high castes’. Berreman related this to the common finding in ethnographic work in the US South (e.g. Davis et al. 2022[1941]; Du Bois, 1898b) where Black people, despite disagreeing with their status position, would still engage in performative strategies in public to avoid white racial violence. Berreman’s (1967b, p. 286) own example of this relates to the US South prior to WW2, whereby: ‘it was not uncommon for Negroes who possessed automobiles to wear chauffeur’s caps when driving them in some southern cities to avoid being attacked by resentful whites to whom the Negro’s low status was inconsistent with ownership of an automobile’.
Put plainly to Cox, therefore, one cannot methodologically jump from observing a group’s obedience towards interactional rules to assuming that group’s uncritical acceptance of a total caste system. As Berreman (1967b, p. 298) summarises, ‘behavioral conformity’ cannot be ‘mistaken for consensus’. The whole point of caste systems, whether in India or elsewhere, is that violation of caste-based interactional rules comes with some degree of penalty – observance of interactional rules can thus easily be construed as strategy for avoiding caste violence.
Lastly, in terms of the slippage between ‘ideal’ definitions of caste in India versus empirical realities elsewhere, we see a fallacious argument in the way that caste is seen in its ideal iteration as being fundamental to Indian society, whereas the principle of caste is seen to be contradictory to the conceptualisation of democracy in the US. Cox (1959) espouses this argument, but it is also seen in the works of those who are more sympathetic to the race-caste approach, such as Johnson (1936, 1944). For Johnson (1945, p. 346), unlike in India, the principles of American democracy – which hold true across the South as much as the North – make the imposition of a ‘true’ caste system impossible to maintain in the US; as he states: A true caste system, however, is based on the acceptance by each individual of his place in the system, which is rigid and not subject to change. This pattern is ruled out in the American South by the basic democratic philosophy of the American creed, to which the South as well as the North adheres in principle. This creed has power enough in the South to make it impossible to exclude the Negro altogether from opportunities for education and self-advancement.
Johnson’s comment is parallelled by Cox (1945, p. 367), who likewise argues that caste violates US creeds while it is the very basis for Indian social organisation: The difference between the racial attitudes of whites and the caste attitude so far as the social ideals of each system are concerned, is that whites wrongfully take the position of excluding groups from participating freely in the common culture, while castes rightfully exclude outsiders from participating in their dharma.
Here, Johnson and Cox conflate US ideal conceptions of democracy with people’s practical understandings of democracy. There is no empirical evidence to suggest that there was a general sentiment among white US Southerners that Jim Crow, for instance, constituted a violation of the American creed. As Spiro (1951, p. 34) noted in the 1950s: Actually[,] discrimination against the Negro is not in violation of southern ideal norms; it is in conformity with them. The ideal norm of the southern American (and of many northerners, as well) is one of inequality, and his behavior is in accordance with his norm.
As Du Bois (2014[1935], p. 109) argued in Black Reconstruction, the US has always struggled with the paradox of being ‘a democracy founded on slavery’. Indeed, empirical research from figures such as Davis (1945) and Wells-Barnett (2013[1970]) have highlighted how state institutions created to maintain democratic rule – such as the courts and police – regularly worked to maintain strict caste status in the US South. In Wells-Barnett’s (2013[1970], p. 51) case, we get an account of how, in 1892, after the murder of two Black grocery store owners in the Curve District of Memphis, the judge of the criminal court ordered the Sheriff to take a 100 men to the peaceful protest and ‘shoot down on sight any Negro who appears to be making trouble’. In Davis’s (1945, p. 11) research, he finds that in the Old and Rural counties of the Jim Crow South, ‘Murder of a white by a Negro is always punished by death, whereas the murder of a Negro by a white is seldom punished by the courts’, while a Black person murdering another Black person likewise is often unpunished by the courts.
The empirical application of democracy in the US, therefore, is explicitly racialised. While scholars such as Cox and Johnson have argued that a caste system violates the (ideal) principles of American democracy, theorists of caste have highlighted how the racialisation of democracy in the US is precisely an example of caste itself. As an aside, both Johnson and Cox would now need to reckon with the fact that – as noted by Berreman (1967a) – the Indian constitution likewise recognises equality as a universal right in its constitution, so even in India there is a discrepancy between concurrent commitments to democracy and caste.
On genesis, legitimation, and function
Besides inconsistent analytical slippages, critics of the race-caste approach also equate the genesis, legitimation, and function of caste in specific contexts with a universal understanding of caste. Such critics have relied upon the separate origins of caste in South Asia, and race in the US, to make the argument that they are incomparable. Parallel to this, such critics have then argued that due to their separate origins, caste and race have been rationalised in different manners, again therefore arguing that race and caste are essentially different social phenomena.
For Cox (1945), the origins of race in the US are capitalist social relations, where race is thus a modern invention; by contrast, Cox argues caste has been traditional to India far before the modern period, instigated by social and religious mores. As Cox argues, ‘modern race relations developed out of the imperialistic practices of capitalism’ (Cox, 1944b, p. 463), that ‘Brahmanism and capitalism are two distinct forms of social organization’ (Cox, 1945, p. 360), and thus race and caste are analytically distinct phenomena. The fact that both caste in India and race in the US then get rationalised in different ways, Cox argues, is further testament to their inherent difference. Race, Cox (1944a) argues, gets defined and rationalised through reference to a (pseudo-scientific) colour-coded biology, whereas caste gets defined and rationalised through religious scripture.
More recently, Pal (2025) engaged in a comparative study of the origins of caste in Bengal and race in the United States in the 18th and 19th centuries to bolster Cox’s argument. Like Cox, Pal (2025) argues that race in the US has its origins in pseudo-scientific understandings of there being inherently different biological ‘types’ of people (i.e. races), that emerge from different species; by contrast, Brahminical concepts of caste start from the supposition that castes derive from the same source. While this pseudo-scientific theory of race thus became instigated into US law – seen through multiple states outlawing interracial marriage, and in school segregation – Pal (2025) argues that the caste system in India was more flexible in the way it advocated for regulations around intermarriage and intercaste mixture (e.g. in the use of shared educational and religious institutions). 3 Due to these supposed differences in origins, Pal (2025, p. 63) thus argues: ‘These two systems, though superficially similar, operated on completely divergent principles; their reckless conflation therefore, is an affront to sociological rigor’.
In making these arguments, both Pal and Cox overlook the examples of how even in India, at times caste has been rationalised biologically through reference to blood-lineage, while often at times race has been defined non-biologically. Mukharji (2014), for instance, has highlighted how natural scientists, such as the Anglo-American Eileen Macfarlane (Erlanson Macfarlane, 1938; Macfarlane, 1936; Macfarlane and Sarkar, 1941), defined different caste groups in biological terms in ways which resonate in contemporary bio-genetic research. As Erlanson Macfarlane (1938, p. 226) noted: Each caste is therefore a biological strain differing more or less from the others. For a true picture of Indian blood-group distribution adequate unmixed samples from any of the numerous castes and tribes are needed from each province.
Interested in genetics and blood, Macfarlane likewise proposed that the genetic composition of the lower-castes of Northeast India was responsible for the overrepresentation of ‘B-type’ blood among the Indian population: That [blood type] B has been in India for millennia and may have originated from the ancestors of the lower castes of the north-east, where the highest concentrations are found now, whence it has diffused into the higher castes. The amounts of [blood type] O and B vary inversely; therefore there might be genes for O in these low caste people with a relatively high mutation rate to B (Erlanson Macfarlane, 1938: 236).
Furthermore, as Shankar (2023) argues, social scientists and political figures of the early-mid 20th century were likewise keen to define caste in biological ways. Govind Sadashiv Ghurye (1969[1932]), for instance, used ‘race science’ metrics – such as measurements of crania, nasal cavities, and phenotype – to ‘prove’ that Brahmins were biologically connected to Aryans. This was a view also endorsed by the British in India in the 19th and 20th centuries, whereby British colonial officials would rank the different caste groups in terms of their blood purity – with the Brahmins being seen as having the ‘purest’ connexion to the Aryans (Thapar, 1996).
Parallel to these biological definitions of caste, we often see non-biological means of categorising ‘race’. Consider, for instance, Morning’s (2015) comparative analysis of national censuses in 2000; here, she found that many national censuses had a slippage between race and nationality, showing both biological and non-biological understandings of race. For example, as Morning (2015) argues, the US had ‘Black’ and ‘white’ as options for one’s race on the census alongside more nationality-based options of descent such as ‘Japanese’ and ‘Korean’; the Solomon Islands explicitly asked: ‘Ethnicity. What race do you belong to? Melanesian, Polynesian, Micronesian, Chinese, European, other or mixed?’; and in Bermuda, one could answer ‘Asian’ to race and/or to nationality questions.
In sum, therefore, Pal (2025) and Cox’s (1945) argument with regards to the biological genesis of race versus the ‘cultural’ genesis of caste seems to be inaccurate. In many cases, caste was rationalised on biological grounds, just as in many cases racial categorisation has not been wholly biologised. However, even if Pal and Cox were correct in their assertion of differing origins of race and caste, they are still making a methodological error in assuming that just because two things have different origins, or because two things get rationalised in different ways, they are beyond comparison.
The fact of the matter, as many comparative historical sociologists would acclaim, is that is that different social phenomena can often have the similar origins, much in the same way that similar social phenomena can have different origins. Magubane (2004), for instance, highlights how the construction of race in 19th century South Africa – where Blackness was construed to be deviant and hypersexual – was constructed by the same Victorian British imperialists who sought to domesticate women within Britian: gender roles in Britain thus had similar origins to racial schemas in South Africa, but they obviously refer to different (albeit interconnected) social phenomena. Likewise, state formation can both have its origins in imperialism (such as with Britain and the US (Go, 2011)) just as much as it can have its origins in anticolonialism (such as with Ghana (Getachew, 2019)) – but this doesn’t detract from the fact that we still see the same social phenomenon of state formation.
With regards to the different rationalisations of race and caste, again we should not assume that different rationalisations automatically mean that we are dealing with incomparable social phenomena. Few would deny that the British policy of ‘indirect rule’ (such as practiced in India) constituted a key component of their colonial apparatus, but we also know that Britain did not apply that mechanism of indirect rule universally across their empire (Go, 2011). Nevertheless, despite their different rationalisations and mechanisms for colonial control, few would deny that British ‘indirect’ and ‘direct’ rule were related to one another and can be subject to comparative research (Lange, 2005). Likewise with caste, one should not be surprised that it gets rationalised differently across different geographical contexts, and these regional differences do not equate to the incomparability of caste. For instance, Barth’s (1960) work in Swat, North Pakistan, found that both Muslim and Hindu communities believed in and reproduced the same caste hierarchy, but that while Hindus used the criteria of purity and pollution to justify this system, Muslims used different concepts of privilege and shame; while the justifications differed, the social structure of caste was the same, and the associated caste behaviours were the same. Berreman (1960, p. 122) makes a similar argument with regards to race relations in the US and caste in India, arguing that despite their different origins, in both cases you see the same functioning logic: The essential similarity lies in the fact that the function of the rules in both cases is to maintain the caste system with institutionalized inequality as its fundamental feature. In the United States, color is a conspicuous mark of caste, while in India there are complex religious features which do not appear in America, but in both cases dwelling area, occupation, place of worship, and cultural behavior, and so on, are important symbols associated with caste status. The crucial fact is that caste status is determined, and therefore the systems are perpetuated, by birth: membership in them is ascribed and unalterable. Individuals in low castes are considered inherently inferior and are relegated to a disadvantaged position, regardless of their behavior.
In sum, there are strong reasons to be sceptical of the critiques of those who have denied comparability between race and caste. Indeed, as I will now expand upon, I contend that often these critics misrepresent, or elide, central race-caste scholarship.
From criticism(s) to a reconstruction: what the race-caste approach does, and does not, do
The fact of the matter is that often the critics of the race-caste approach do not sufficiently engage with the corpus of scholarship developing this said approach. While Pal (2025, p. 63) comments that comparing race and caste is a ‘terrible idea’ which ‘is an affront to sociological rigor’, the only paper they engage with in favour of the race-caste approach is Warner’s (1936) American Caste and Class – Pal thus ignores a wide range of empirical and conceptual scholarship on the matter. Likewise, while Cox (1959) does make more reference to the empirical scholarship on race-caste, he does not necessarily engage with their central arguments; for instance, Cox’s (1959) assertion that the race-caste approach has no hypotheses is erroneous, as we have already alluded to (in terms of Davis’ theory of caste and violence), and as we will refer to throughout the rest of this paper. Likewise, Cox’s (1942) assertion that the caste school are essentially substituting the concept of ‘race’ with ‘caste’ is also erroneous, significantly misreading the intentions of said authors and their application of the caste concept – again as I will outline below.
The matter is complicated by the fact that even those writing in support of the race-caste comparison do not always do so with full reference to the field. Holmwood (2024), who is sympathetic to the approach, for instance, lumps Franklin Frazier and Charles S Johnson into the race-caste school without outlining their research programmes. Both Frazier and Johnson had an equivocal relation with the concept of caste, to the extent that Cox (1942) jokes that Johnson only toys with the race-caste approach rather than fully supporting it. Indeed, as I noted earlier, at times Johnson (1945) was highly critical of the concept of caste as it applies to the US, whereas at other times – as I shall outline below – he saw its empirical viability as a concept that can be used to describe a specific iteration of race relations. Likewise with Frazier, at times he uses the concept of caste (Frazier, 1957) and at other times – such as in his review of Black Metropolis – he explicitly denies the utility of the caste concept in the US (Frazier, 1946). Suraj Milind Yengde (2024) has recently written an excellent paper in defence of the race-caste approach, but even then his focus is more so on a handful of caste authors which excludes some key figures (e.g. the aforementioned Charles S Johnson and Franklin Frazier, and also St Clair Drake) and does not – in virtue of the scope of the paper – delve either into the empirical cases highlighted by the race-caste approach, nor on their tentative hypotheses and conceptual frameworks. 4
With this in mind, I wish to spend the remainder of the paper outlining what the race-caste approach scholars were actually arguing. My argument is that the race-caste scholars of the classical era developed a sociological programme oriented not towards demonstrating an equivalence between race and caste, but rather oriented towards the relational empirical question of when, how, and why, in certain specific moments, racial domination becomes caste-like. Furthermore, these scholars demonstrated that the answer as to when, how, and why racial domination can become caste-like cannot be answered a priori but must be ascertained empirically; scholars used the notion of caste precisely to show how it offers an empirical continuum to judge the rigidity of a given system of racial domination.
The historical specificity of the race-caste approach
Very few – if any – of the race-caste approach scholars were brazenly arguing that race is directly equivalent to caste. Warner and Davis (1939, pp. 232–233; emphasis added), for instance, emphasise in their defence of the race-caste approach that: There has been no attempt [. . .] to demonstrate that our caste structure and Indian caste structure are exactly the same, but rather we have attempted to show that they are the same kind of social phenomena and, therefore, for the comparative sociologists [. . .] they are forms of behavior which must have the same term applied to them.
For Warner and Davis (1939), ‘caste’ and ‘race’ are both types of rank, much in the way that lemons and limes are both types of citrous fruits. They argue that while substantially distinct, race and caste can at times be compared precisely in virtue of the fact that they broadly belong to the same analytical space. 5 This means that both Warner and Davis admit that there is no mirror image between caste in India and race in the US South. In A Comparative Study of American caste, Warner and Davis (1939, p. 233) thus note that the extent to which race and caste can be compared across the two cases is an empirical question to be tested rather than a conceptual claim that can be assumed; as they state: the adequacy of the race-caste comparison relies on the ‘extent to which it fits the various types of relationships between whites and Negroes in this society’.
Warner and Davis, among others, thus argued that the race-caste approach was one generated in order to guide empirical researchers to consider the rigidity of specific systems of racial domination in the US. The race-caste approach was generated from empirical research, often in the US South, where researchers were consistently finding that racial domination often appeared to be rigidly caste-like in the way that the racial system was, to come back to Wacquant’s (2024, p. 271) earlier definition of caste, ‘constructed by (i) a rank-ordering of (ii) birth-ascribed, descent-based groupings that are (iii) endogamous, in which (iv) membership is permanent and transmissible, ands (v) whose hierarchical separation is justified by a collective belief in purity or congenital superiority’. Importantly, this means that we must note that the race-caste approach was one deliberately generated from specific case studies to both describe and explain certain social relations (often in the US South).
The historical specificity of the race-caste approach is highlighted both by Du Bois (2014[1935]) and Frazier (1957). Indeed, while Frazier (1946) disagrees with the use of ‘caste’ to describe race relations across the whole US, he understands the utility of the caste concept to ‘represent both a stage beyond slavery in the development of race relations and also a generic form of social organization in which people with different racial backgrounds find accommodation’ (Frazier, 1957: 253). It was in this period in the aftermath of enslavement that Du Bois (1898a, p. 15) discusses the move from a ‘caste of condition’ (i.e. enslavement) to a ‘caste of race’. 6 For Du Bois, it was in this era of abolition that US Southern states used their legal apparatus to instigate Black codes, Jim Crow segregation, and endogamy, with the support of educational and economic institutions, to create a two-tiered caste system between the superior (whites) and inferior (Black people). With a newly imposed system of debt peonage, along with explicit political disenfranchisement, Du Bois argued the in the US South ‘the black masses trapped by caste were defenseless neo-slaves within a new system of oppression’ (Morris, 2015: 39). Du Bois’s analysis parallels Frazier’s (1957, p. 259) argument that the post-enslavement era involved ‘a legalized caste system [. . .] accomplished by reducing public education for Negroes, by disfranchising them politically, and by instituting a system of racial segregation or Jim Crow’. Indeed, speaking particularly about the exclusion of Black workers in the US South from new lines of work in the late 19th century – particularly the textile industry – Frazier (1957, p. 259) goes as far as to say that ‘caste based upon race has been the most important feature of the occupational structure of the South’. Importantly, Frazier – following Du Bois – saw this caste system in the US South as explicitly being based upon race, thus emphasising that this group of scholars were not equating the two concepts.
Throughout his historical excavation of ‘caste based upon race’ in the US South, Frazier (1957) regularly makes positive reference to the empirical work conducted by Allison Davis and John Dollard. It was through these studies – Davis’s (2022[1941]) co-authored Deep South, Dollard’s (1988[1937]) Caste and Class in a Southern Town, and Davis and Dollard’s (1964[1940]) Children of Bondage – that the race-caste approach was empirically bolstered. Deep South and Caste and Class both drew upon research conducted in Mississippi, in Natchez and Indianola respectively, while Children of Bondage was comparative across Natchez and New Orleans. In all cases, the authors argued that the system of racial domination was rigid to the extent that it was comparable to caste.
In Deep South, caste is shown and proven, not assumed. Warner (2015[1945]) – an advisor behind the Deep South study – states this in a Methodological Note to Black Metropolis. As Warner (2015[1945], p. 774) notes, Davis and his team went to Natchez
7
with a driving research question: ‘What is the type of social system in Deep South which organizes the lives of Negroes into a subordinate level and of whites into a superordinate level?’; in answering this question, they found that the ‘class hypothesis’ was insufficient, which drew them towards theorising the caste comparison: Simple tests of the class hypothesis easily convinced us it did not fit the facts. Negroes and whites cannot intermarry; members of two classes can. Negroes cannot rise out of the lower to the higher white levels; members of a lower white class are able to do this. Unlike the ethnic individual, the Negro individual’s symbols of inferiority cannot be removed; they stay with him until death.
Davis and his team thus moved from the empirical realities of Natchez to the overall conceptual framing of caste, and not vice versa. In terms of these empirical realities, which I will further expand upon later in this paper, his team found a caste system both in the urban (termed ‘Old City’) and more rural (‘Old County’) areas of their research site. This system was evidenced by:
i)
ii)
In India, linguistic forms for addressing low caste people are often the same as those for addressing children, and they are treated as children. In colonial states it is standard to speak of and treat the “natives” as children or childlike and to foster a kind of dependent role among them analogous to the child’s role in the family. In the southern United States, Negroes are called by their first names as are children or are addressed as “boy” regardless of age. In Mounds State Park in Alabama I have seen restrooms for whites labeled “Men” and “Women” while those for Negroes were labeled “Boys” and “Girls”.
Beyond infantilisation, Davis et al. (2022[1941]) found that other white participants would rationalise the caste system through reference to religion. As one of their participants, a physician, noted: God didn’t put the different races here to all mix and mingle [. . .] I don’t say He put the Caucasians here to rule the world or anything like that [. . .] but since they have a superior intellect and intelligence, I don’t think God would want them to mingle with inferior races and lose that superiority. You know the Negro race is inferior mentally, everyone knows that, and I don’t think God meant for a superior race like the whites to blend with an inferior race and become mediocre [. . .] I think I am right in saying that, and my attitude is Christian-like. (Davis et al., 2022[1941]: 13–14).
Again, the findings here from Deep South resonate more broadly with the race-caste approach scholarship. Berreman (1967b, pp. 295–296), for instance, argues that references to religion to justify the caste system are expressed just as much in the US South as they are in the Indian context, declaring: ‘In the United States the southern racist quotes scriptures to justify his position just as in India the Brahman quotes scriptures to justify his position’.
iii)
iv)
Not only is there a formal law which prohibits intermarriage, but the society also seeks to prohibit any sexual intercourse outside of marriage between white women and colored men. Any Negro man who makes advances toward a white woman, even though she be a professional prostitute, has broken the strongest taboo of the system and risks a terrible punishment [. . .] the Negro man is thought by the whites to be always a potential rapist [. . .] So ever present is this fear, that Negro men are often afraid to approach strange white houses where there may be a woman alone, or even to approach too near a lone white woman on the streets at night.
Importantly, these caste sanctions relate to marriage and Black men/white women relations in specific. There is not a strict taboo, for instance, on white men having sex with Black women; however, in the case a white man and Black woman have a child, the child is racialised as Black, and is expected to abide by the caste rules as a Black person. Indeed, here Davis and his team’s finding replicates Dollard’s (1988[1937]) own research in Indianola, where he notes that despite the strict taboo on Black men/white women sexual relations, white men get allocated ‘sexual gains’ through the caste system which permits them to have relations with Black women.
I have written about Deep South at length because it precisely shows the logic of the race-caste approach. Allison Davis and his colleagues did not go into Natchez knowing that a caste system was going to be present. Rather, as Drake (1974) – one of the Deep South team – recounts, Deep South was hypothesis driven. Drake (1974) notes that Warner and Davis originally wanted to find a point of comparison to see how race relations worked different across the US North (which they studied in Newburyport 8 ) and South – Natchez became their Southern case study. Knowing that class relations were fundamental to race relations in Newburyport, Warner and Davis sought to work out whether the US South followed the same pattern, or whether it followed a different model of hierarchisation; their goal in Natchez thus became ‘explanatory, understanding how that particular society worked and testing the caste-class hypothesis with the idea that ascertaining the facts would have some relevance for change’ (Drake, 1974: 48). That race relations in the US South were ‘caste-like’ was thus a research finding, rather than a conceptual assumption that had been made prior to the establishment of facts. Race relations as being caste-like enabled Davis and his team to not just describe, but also explain social phenomena: from the interactional rules between white and Black people, through to segregation, the economy, and social, political, and religious organisation.
Importantly, in Deep South as well as elsewhere, scholars of the race-caste approach emphasised that race relations were caste-like, not that they were a mirror image of caste patterns seen elsewhere. This leads to my second substantive point: scholars of the race-caste approach understood ‘caste’ as a continuum, through which one could ascertain the given rigidity of a given system of racial domination.
Caste as continuum of racial domination
Especially in the early-mid 20th century, where much of the race-caste scholarship emerged, most race scholarship happened under the banner of ‘race relations’. However, ‘race relations’ is a broad term, merely summarising – as it says on the tin – the systematised relations between different racialised groups (Frazier, 1947). The race-caste scholars thus used the notion of caste to highlight a specific constellation of race relations – or racial domination more broadly – referring to those instances where the social, political, religious, sexual, economic, educational, and cultural organisation of society was most rigidly organised along racial lines (as we saw in the above example of Deep South).
In thinking about caste as a possible arrangement or ‘type’ of racial domination, scholars of the race-caste approach are not arguing that race relations must be identical in all measures to the South Asian application of caste. Rather, in highlighting caste as a continuum, such scholars are showing that there will be times where the US case may be ‘looser’ than the South Asian application of caste, and vice versa – this does not mean that either case is not caste-like, but rather highlights how you can have different degrees of rigidity in social hierarchies. 9
Again, this point about caste as a continuum is repeatedly made by classical race-caste scholars. Berreman (1960, p. 121), for instance, describes this very explicitly when he states that caste, in social scientific literature, is: [. . .] describing an ideal type at one end of a continuum along which systems of social stratification might be ranged. There can be little doubt that the systems in India and the southern United States would fall far toward the caste extreme of the continuum.
Berreman (1967b, p. 302) repeats this point consistently throughout his work, stating that: [. . .] we have a continuum from caste to non-caste organization and it is characterized by a continuum in the manifestation of the various attributes of caste organization.
Classical authors of the race-caste approach thus argued that thinking about ‘caste’ as a specific point along a continuum offered them empirical avenues into the study of race relations, allowing the empirical measurement of how rigid and ritualised a given structure of racial domination is. This is precisely why, as aforementioned, Davis and Warner used Natchez as a space where they could empirically test the looseness (class hypothesis) or rigidity (caste hypothesis) of Southern race relations. Indeed, social scientists – such as Davis (1945) and Drake (1955) – use this approach to detail the extent to which different aspects of the total social space (e.g. the economic, social, political, sexual, religious etc.) are ordered across the caste continuum.
For instance, Davis (1945) highlighted in his studies of urban and rural Mississippi that the economic component of society, contrasted to the political, social, and religious components, was less rigidly organised along caste lines. In the political sphere, Davis (1945) argues, caste segregation is clearly maintained by the exclusion of Black political figures and voters. Recounting the research from Deep South, Davis (1945, p. 11) comments: Only six Negroes in Old City had been allowed to register as voters in national elections; these registered Negroes admitted they did not vote, because they felt, as a leading Negro professional man said, that government was “the white man’s business.” In the rural counties, no Negro was registered. No Negroes were registered as voters in state or municipal elections, nor had any been a candidate for any county or municipal office since Reconstruction. The white officers in charge of registration and elections stated to one of the white interviewers that the whites prevented Negroes from registering as voters by intimidation and, if necessary, by violence.
For Davis, therefore, it seems viable to argue that the political structure of the Deep South was rigidly organised along caste lines – even in the urban Old City. This is contrasted to the economic structure of the Old City, where free competition, especially among white and Black workers, meant that the economic structure was not so rigidly organised along caste lines to the same extent as the political (though caste was still maintained to a looser degree). As Davis (1945, p. 12) argues ‘although color-caste sanctions operate as occupational taboos to a high degree, the economic structure is not caste-segmented upon an all-or-none color basis, as are the political and religious structures’. This relative ‘caste looseness’ can be seen in various ways. For instance, Davis (1945, p. 12) highlights that in ‘storekeeping, contracting, farming and professional service to colored persons’, white and Black workers are in competition with one another to the extent that such occupations are not wholly segregated along racial (cum caste) lines (i.e. they are not clearly divided into ‘Black’ and ‘white’ jobs). Indeed, because of this free competition you even see the development of a small fraction of upper-class lower-caste groups (i.e. a Black bourgeoisie), which again shows that: [. . .] the economic system [. . .] has thus prevented the full development of caste -that development in which all members of the lower caste are legally, or by virtue of unbreakable custom, below all members of the upper caste in wages, occupational status, and the value of property owned (Davis, 1945: 2).
However, as Davis (1945) points out – again demonstrating the ‘caste as continuum’ approach – parallel to this free competition dynamic, certain remnants of caste organisation still shape the economy. For instance, due to the caste belief in the inferiority of Black people, white capitalists in the Old City – rather than simply excluding Black workers en masse – sometimes prefer to hire Black labour (on cheaper wages than whites), and rent their land to Black people (for higher levels of rent). Davis (1945: p.14) describes this as involving the ‘modification of the caste system in the interests of the profits of the upper and middle economic groups of white people’. Importantly, as Davis (1945) clarifies, such modifications which ‘loosen’ the caste organisation of the economy are only permitted if the overall position of the dominant caste remains stable. It is for this reason, Davis (1945: p.14) argues, that you do not tend to see Black people hired by whites into white collar occupations – this does not serve the interests of the white capitalist class: A white store which employed colored clerks, for example, would be boycotted by both these groups. The taboo upon the employment of colored workers in such fields is the result of the political power and the purchasing power of the white middle and lower groups. As a result of these taboos in the field of “white-collar” work, the educated colored person occupies a well-nigh hopeless position in Old County.
This ‘caste as continuum’ approach was also endorsed by one of Allison Davis’ mentees: St Clair Drake. Drake came to the U.K in 1947 to conduct research for his doctoral project titled Value Systems, Social Structure and Race Relations in the British Isles. 10 Under the guidance of Lloyd Warner, and having worked on both Deep South and Black Metropolis, Drake’s PhD project intended to use the neighbourhood of Tiger Bay in Cardiff (Wales) as a point of comparison to race relations in the American South and North. Summarising his PhD, Drake (n.d.) thus comments that his intention was to study race relations in Tiger Bay, positioning such relations along the class-caste continuum.
Drake’s argument in the dissertation is that the caste system in Tiger Bay progressively loosened from the early 20th century to the 1940s. Between 1919 and 1935, Drake argues that the caste system was at its most rigid. The National Union of Seamen, for instance, feared competition from Black and Brown workers, culminating in the state passing the 1925 Coloured Alien Seamen’s act, which heavily restricted the representation of Black and Brown workers in this industry. Coupled with this economic exclusion, the 1925 Act also put more stringent conditions for citizenship on ‘coloured’ workers, resulting in many Black and Brown workers having to register as ‘aliens’; given that the spouses of ‘aliens’, through the 1914 Nationality act, would also become ‘aliens’ through marriage, the 1925 act was also construed by Drake as putting a legal taboo on interracial marriage. Indeed, on this theme of sexual relations, Drake points out that mixed-race children were consistently treated as ‘lower caste’, often resulting in mass unemployment among mixed-race (described as ‘half-caste’) people through the early 1930s. Further, it is in this period, Drake (n.d.) argues, that churches became segregated along caste lines in Tiger Bay, while white people either started leaving the neighbourhood en masse, or created their own white enclaves within the neighbourhood, creating a caste-like residential and social segregation not dissimilar to the urban realities researched in Deep South. Drake contrasts this period of 1919–1935 with the 1935–1947 period, in which he argues that race relations became less caste-like and more organised along the axis of class. Here, pre and postwar attitudes among Brits, Drake argues, made them more liberal towards interracial mixture, employment needs and an increasingly active labour movement meant that occupational segregation became less common, and the increasing presence of people of colour – through migration of military personnel as well as postwar migration – also meant that educational institutions around Tiger Bay became desegregated while interracial sexual relations became more normal.
For the sake of this paper, I have referenced Drake’s work alongside Davis’s because they both argued that there were scientific gains from thinking about caste as a continuum to empirically and comparatively test the rigidity of the given system of racial domination. For Davis (1945), a key comparison was between different levels of society. He was questioning how the economic level of society, relative to the political, sexual, social, cultural etc., was structured along the lines of caste, coming to the argument that while the economic level certainly was still organised by caste, it was more ‘loosely’ organised than other levels of society. With Drake, his comparison was temporal: he examined how, in the case of Tiger Bay, race relations never became fully crystallised into a cohesive caste system, and that the race relations thus began towards the rigid caste-side of a continuum and gradually became looser. In both cases, we see Davis and Drake think of caste not as a direct equivalent to race, but rather as a continuum; in thinking of caste as a continuum, both Drake and Davis were able to empirically measure and explain race relations in particular historical moments.
Indeed, thinking of caste as a continuum was a dominant approach among the classical advocates of the race-caste comparison. For such scholars, there were two main scientific benefits of thinking of caste as a continuum for measuring race relations, both of which I will now review in this paper. First, as previously mentioned, scholars argued that thinking of caste as a continuum could help to develop comparative insights into hierarchical relations across time and space, and secondly, they argued it could help develop insights into mismatches between beliefs, practices, and social structure.
The race-caste approach and comparative sociology
Scholars used the race-caste approach to develop comparative insights across space and time. While history of sociology textbooks tend to portray the early Chicago school as being primarily micro-ethnographic, we must not forget the members involved in this putative tradition were primarily concerned with sociological comparison – whether that be Warner comparing the class-caste hypothesis in Newburyport, Natchez, and Chicago, Drake comparing Bronzeville, Chicago to Tiger Bay, Cardiff (or indeed, Tiger Bay to other U.K cities such as Hull and Liverpool), or Allison Davis contrasting the rural and urban South. These scholars of race-caste all embodied the ethos later declared by Berreman (1967, p. 351) that: ‘scientific endeavor requires comparison’. For these social scientists, viewing caste as a continuum opened up their possibility ‘for travelling across the span of historical [and geographical] cases and fostering rigorous comparison liable to nourish further theorizing’ (Wacquant, 2022: 77).
Consider the work of Allison Davis, and especially his work in Deep South. As already mentioned, Deep South itself was figured as a comparative project where a Southern case could be contrasted with a Northern one (Drake, 1974). Indeed, Deep South was also intended as a contrasted case study to Drake and Cayton’s (2015[1945]) latter study of Bronzeville, Chicago, in Black Metropolis. In fact, Warner’s (2015[1945]: p.781) ‘methodological note’ in the Appendix of Black Metropolis has a table with a series of compared measurements between caste and race relations between Bronzeville and Natchez, which he uses as the basis to argue that even in Chicago: the type of status relations controlling Negroes and whites remains the same and continues to keep the Negro in an inferior and restricted position. He cannot climb into the higher group although he can climb higher in his own group. Legally, he is permitted to marry across the color line but there is very little intermarriage. The children of such marriages are always negro and suffer, as do their parents, the ‘restrictions’ and deprivations of the Negro caste. The rewards and punishments, the rights and duties, knowledges and advantages are unequally distributed. In short, there is still a status system of the caste type.
Importantly, however Deep South was not only used as a comparative case with other cases (e.g. Newburyport and Bronzeville), but it was already a comparative study between the urban and rural. Davis and his team, through studying the rural Old County versus the urban Old City, were able to demonstrate the different levels of the ‘caste like’ organisation of race relations. For example, in the rural area, there is a clear organisation of the (plantation) economy across caste lines, which in turn creates a visibly clear hierarchy between white and Black people, which is then supported through social institutions (e.g. education). This creates a situation where ‘almost all the colored people are families of poverty-stricken tenants and almost all the white people are families of owners of large landlords, [and thus] caste is almost “perfect” economically and socially’ (Davis, 1945: 15). Indeed, in the rural Old County, this caste organisation is so ‘neat’ in the way the political, economic, and social align that, for instance, there was large-scale resistance among rich whites to improve the provision of schooling for Black children. As Davis et al. (2022[1941], p. 191) detail this relationship between the economic and social fitting of caste: Many landlords and some members of the county school boards were opposed to the development of more efficient schools for rural colored children. They expressed the fear that the providing of even a thorough grammar school education for colored students would make them unwilling to remain on the plantation and would end by depleting the supply of workers. Equally important in many cases was the belief that educated colored people were less amenable to the caste sanctions, less deferential, submissive, and dependent, and therefore a danger to the efficient working of the caste system.
Davis et al. contrast this rural Old County with urban Natchez, where you have – to an extent – a gradually diversifying workforce given the burgeoning capitalist, industrial economy. In this urban setting, Davis et al. (2022[1941], p. 18) argue, while the economic organisation becomes less clearly structured along caste-lines, whites in the Old City still deeply hold caste beliefs, especially with regards to the notion of a ‘properly deferential’ Black person, whereby: A properly deferential Negro is a “good” Negro, and the highest praise offered is that a Negro is “humble and has the proper attitude toward the whites,” that “he knows his place.” On the other hand, the Negro who does not abide by these rules, who is not properly humble and deferential in his behavior, is “sassy,” “uppity,” and “bad,” and is trying to “get out of his place.”
While deference is also expected in the rural Old County, Davis et al. (2022[1941], pp. 215–216) highlight how deference takes on an increasingly important symbolic value in the Old City. In the Old City, because – unlike the Old County – many ‘white men [. . .] are almost as poor as the colored workers and occupy approximately the same occupational level’, both white workers and capitalists work even harder to ‘impress the colored group with the fact that economic equality [. . .] is not real equality [. . .] that caste exists all along the line, as the myth demands, and that actually any white man [. . .] must be treated with the appropriate deference’. Indeed, this also explains why, as Davis et al. (2022[1941]) recount, in the Old City many whites lament the fact that Black people do not know their place as well as those Black people do in the Old County.
Importantly, these comparisons between the rural and urban in Deep South allowed for Davis et al. to build tentative hypotheses that could then help build sociological theory which could be tested in further empirical research. For instance, Davis (1945) uses the research from Deep South, as aforementioned, to generate a hypothesis as to when violence is more or less likely to occur in a caste-like organisation of race relations. Here, Davis (1945, p. 15) argues that when caste-organisation is mostly rigidly sustained, violence tends to be less often-occurring, whereas ‘when the castes are in economic competition as laborers and tenants, however, violence and conflict seem to be at their height’. Deep South, therefore, offers Davis (1945, p. 7) the empirical material – itself comparative by nature between the urban and rural – to ‘state a theory of violence as a reaction to the breakdown of caste in the economic sphere’. Comparison helps Davis generate hypotheses, which helps him generate theory.
This comparative-hypotheses-theory generating model was also foundational to Drake’s work. Indeed, the comparative inclination was explicitly formed in Drake’s PhD project on Tiger Bay, Cardiff, where he sought to examine the constellation of race relations into a caste-like system in comparison to other British cities such as Hull (which he argued was looser) and Liverpool (which he argued was more rigid). Moreover, in the conclusion of Drake’s dissertation he explicitly calls for a comparative sociology of race and caste. Here, Drake (n.d., p. 15) suggests the potential to compare his own British case study, and Deep South, with colonial territories – such as Mombasa and Nairobi in Kenya – where ‘the British ideological systems operate to moderate caste tendencies’. Such comparative studies, Drake argues, would allow us to examine how race relations can be crystallised (or not) into caste regimes in spaces where people of colour are, or are not, numerical majorities.
Indeed, Drake took it upon himself to engage in such comparative research throughout his career. In Some Observations on Interethnic Conflict as One Type of Intergroup Conflict, for instance, Drake (1957) compares the racial-caste system of the US that he studied in Deep South with colonial case studies such as 1950s Kenya, and South Africa. Starting from the empirical fact that inter-ethnoracial violence was more common in Kenya than South Africa, Drake – like Davis – uses caste as an explanatory framework to generate hypotheses about when violence is more or less likely to occur. Namely, Drake (1957, p. 177; emphasis added) argues that just like in Deep South: There is considerable evidence that within societies organized on the principle of caste, as the proportion of people in the subordinated caste increases relative to the proportion of superordinates, the more rigid become the caste controls [. . .] Thus, throughout the South, the so-called “Black Counties” have had the poorest schools and the greatest measure of discrimination. These are counties where the structural situation begins to approximate that of the African cases.
Drake, like Davis and others, thus goes from the empirics of his case study towards a hypothesis-generating, explanatory framework of caste. As he summarises himself (Drake, 1957: 177), a comparative ‘system analysis’ of race-caste allows him to ‘isolate certain problems within the systems and to state hypotheses that may have validity for all systems of caste, irrespective of the local variations’. The race-caste approach, as we can see in the works of Davis and Drake, therefore, was developed as a hypothesis-driven conceptual framework, generated by empirical research, in which the authors sought to make cross-case comparisons without conflating comparison and equivalence.
Critics have often failed to understand this methodological side of the race-caste approach. For instance, Pal (2025) argues that the stricter regulations on marriage in the US South, compared to the possibility (and frequency) of intercaste marriage in Bengal, demonstrates that race in the Southern US cannot be thought of along the lines of caste. However, the methodological ethos of the race-caste approach shows how such a critique is erroneous. In a comparative lens, one could simply use the same empirical data to make the argument that sexual relations are more caste-like in the US Southern states than in Bengal. Indeed, this is not dissimilar to the argument of Davis (1945), who argues that at times the caste system in the US is more rigid than that in South Asia; likewise, Warner and Davis (1939) make a similar point about the rigidity of caste in sexual relations by pointing out the greater frequency of intercaste marriages compared to interracial marriages. Furthermore, one could go further to again reject Pal’s (2025) argument, instead showing how the Bengal and US South cases of intercaste/interracial marriage, despite their differences in frequency, show the same basic premise: that higher status men are able to mould the caste system to their own benefit. I have already noted the fact that in the US South, while marriage was a taboo, sexual relations between white men with Black women was not (whereas Black men having sexual relations with white women would be grotesquely punished). By parallel, as Pal (2025) shows, it was largely upper caste men who would intermarry with lower caste women, and not vice versa (this is also a point made by Berreman (1967a)). What Pal (2025) sees as a critique of the race-caste approach could thus actually be a good starting point for comparative analysis of race, caste, and gender.
Likewise, Cox (1944a) argues that while different castes in India have strict occupational segregation, the same reality does not exist in the US South. Cox (1942, p. 221) argues that while you may have lower-paying occupations traditionally done by Black folks in the US South, this should not be conflated with the Indian context, where different castes themselves have traditional occupations; as Cox summarises it: ‘every caste has a traditional occupation, and not that every occupation has a caste’. In other words, for Cox, while some jobs in the US may typically have been filled by Black labourers, this does not itself mean that Black labourers themselves had a typical occupation. However, much like with Pal, this line of reasoning is open to critique. As Davis (1945) and Berreman (1967a) argue, one could argue comparatively that the economic levels of society are more rigidly organised along caste lines in one context than the other. Indeed, if one pursued this line of argument, coupled with the aforementioned comparison of intermarriage, one could develop a comparative argument that ‘while sexual relations are more organised along caste lines in the US, economic relations are more organised along caste lines in South Asia’.
Indeed, some authors – such as Johnson (1936) – have relatedly made the argument that while occupational segregation along caste lines may not be as salient in the US compared to South Asia, many people in the US at least believe that Black people do have certain traditional occupations. In this context, Johnson thus points out the necessity to avoid conflating the notion of a ‘caste structure’ with ‘caste beliefs’: sometimes caste beliefs exist in a structure which is not itself fully realised as a caste system. This leads us to our final point: beyond providing comparative insights between hierarchical regimes, scholars of the race-caste approach argued that the concept of ‘caste’ also enabled insights into the mismatches between structure and belief patterns.
Caste-like structures and caste-beliefs: on structure, beliefs, and practice
Potential mismatches between structure and beliefs have been a central topic of sociology, with Bourdieu’s (1977) notion of habitus clivé being a known response to this problem. For Bourdieu, such a clivé occurs when there is a change in the overall material relations of society, but members of that given society may still have internalised belief patterns from a previous era.
Johnson (1936), prior to Bourdieu, directly wrote about such a habitus clivé when discussing the breakdown – and attempted reinstitution – of ‘Negro jobs’ in the US during the Great Depression and WW2. In both these eras, Johnson argues, the economic conditions of the US South (and North) necessitated a diversification of the workforce, but in both cases white workers wanted to avoid the stigma of doing ‘Negro jobs’ in ways which reproduced caste-practices in the labour market, while new forms of industry allowed for white workers to create new hierarchies of moral worth in labour along caste lines.
Consider Johnson’s analysis of the tobacco industry in the US South. During the periods of enslavement and abolition, this industry, Johnson (1936) argues, was typically seen as reserved for Black labourers, and in labelling such occupations as ‘Negro jobs’, the work itself was stigmatised. However, as the Depression took hold in the 1930s, white workers – by necessity of their economic conditions – had to start working in these areas of employment previously stigmatised as Black. This created a new ‘competition for “Negro jobs”, [. . .] affecting the traditional basis of race relations, despite the sanctity of the mores’ (Johnson, 1939: 325). While the mores of society – including dominant belief systems – stuck rigidly to caste-attitudes, the economic conditions necessitated occupational desegregation. However, as Johnson (1936) argues, even in this context caste-beliefs were strong enough to the extent that whites tried to develop new caste lines within traditionally Black jobs which were now desegregated. Thus, in the tobacco industry, Johnson (1936) points out that newly entering white workers tended to work as machine operators or ‘searchers’, while Black people remained as stemmers, janitors, and manual labourers – white workers would then see their own labour as having more moral value, and would redefine ‘Negro jobs’ within the occupation itself, creating new caste lines at the more micro level of the economy in terms of task segregation within an overall occupation. As Johnson (1936, p. 65; emphasis added) writes: Searching follows stemming, and the searchers can check on the stemmers rather than be checked on by the stemmers. This, now, permits it to be classed as a white job. In most places all machine operators and cigarette makers are white. In many cases where the conception of what constitutes a Negro job has been strictly construed, Negro jobs have been reduced to that of janitor, or limited to such heavy work as handling hogsheads. The Negro worker is by tradition to stay in his place, but his place is being continually restricted.
Johnson (1944) expands this argument when viewing the period of WW2, where the US had to increase its industrial production to provide supplies to the Allied forces of Europe. Here, Johnson (1944, p. 587) argues that the increasing racial diversification of the workforce on the one hand meant white workers taking so-called ‘Negro jobs’, and on the other hand, Black labourers entering into previously white-only industrial occupations, in both cases causing a ‘violation of the caste structure’. To an extent, Johnson (1944, p. 587) thus argues, the ‘imperatives of the new technology inherent in our dominant economic system’ created a form of ‘disintegration of the remnants of the caste structure’. However, despite the economic necessity to disintegrate the caste structure, Johnson (1941a) shows that caste beliefs remained strong – just like in the Depression era – which created new moral boundaries in the economy being drawn along caste lines. As Johnson (1941a, p. 588; emphasis added) writes: At a time when it is imperative for industry to increase its production [. . .] both individual industries and organized labor are clinging to their traditional policies founded upon the outmoded caste patterns.
Johnson (1941a, p. 589) points out how these caste practices thus created strict segregation not only in the micro-level of certain industrial occupations, but also in the military itself, stating that: Black volunteers in the military service have been denied except in all Black units [. . .] most of the existing military units open to them are service units; that no Negroes are enlisted in the navy except as mess attendants; that no Negroes are enlisted in the Marine Corps; and none in the active flying service. There are no Negro officers in training, no nurses, no physicians and dentists in the Medical and Dental Corps.
Johnson’s analysis, therefore, teases out the mismatch between people’s attitudes and the economic conditions in which they are living. While he argued that ‘caste’ may be incongruous with the economic basis of society, Johnson points out that the fact that people still had deeply held caste beliefs and attitudes had clear material consequences. For Johnson (1941a), analysing people’s enduring caste beliefs holds more explanatory power than the more porous notion of ‘racial prejudice’: he argues the desire for segregation and hierarchy built into the very definition of caste itself makes it the more scientifically accurate concept to use in the situations he is describing.
Indeed, it is in this conceptual point that we can draw a parallel between Johnson’s argument with Frazier’s. Johnson’s (1936, 1941a) underlying argument – itself similar to Monroe Work’s (1937) – is that as the US South moves from an agricultural, plantation economy to an increasingly urban, industrial, capitalist economy, the caste beliefs become more dissonant with the economic needs of society, and will eventually dissipate. The period of the 1930s and 40s, Johnson argues, is an example of where there is a lag – what Bourdieu would later refer to as a habitus clivé – between the structural relations of society and the beliefs that ‘fit’ these relations. By parallel, Frazier (1957, p. 265) writes that ‘the system of social control which the caste pattern of race relations undertakes to maintain is opposed to the economic and social changes now occurring in the South’. Unlike Johnson, Frazier (1957) goes a step further to then make a comparative argument with caste in India. Frazier (1957) argues that even in India you see how especially caste segregation on public services is decreasing as imperial-capitalist intervention creates new forms of transport to improve the efficiency of the (exploitable) workforce. 11 Like Johnson, Frazier thus contends that eventually dominant caste beliefs may dissipate in order to match the changing social structure of capitalist economic arrangements – both in the US South and in India.
Again, we therefore see how classical authors made use of the race-caste approach as a methodological tool, which they argued improved scientific rigour. Especially Charles Johnson used the race-caste approach to highlight situations where you had a lack of a caste structure, but persistence of caste-beliefs and practices, and indeed therefore attempts to create new patterns of caste segregation in the economy which sometimes were, and sometimes were not realised. In this approach, Johnson highlights that caste beliefs can exist without a solidified caste structure, and therefore that ‘caste’ can be studied (through focussing on attitudes, practices, and beliefs) even in the absence of a realised caste structure. Johnson’s work therefore offers a rebuttal to the criticism articulated by Pal (2025) and Cox (1944a) who argue that caste cannot exist in the US because social relations are not ordered as they are in places where caste is said to exist – such as in South Asia. As Johnson and Frazier, among others, argued, the overall social structure between these cases may not be the same, but one can still emphasise the endurance and effects of caste beliefs.
Furthermore, through focussing on caste beliefs, such scholars also respond to Pal’s (2025) argument that the race-caste comparison does not work because race relations were strictly ordered by state control in a way that does not translate to the practice of caste in Bengal. Pal (2025) is right to emphasise the importance of the state to race relations in the US South, but his analysis does not centre enough on the caste attitudes of those in the South, nor on the non-state sponsored caste practices then built upon from these attitudes. Even in the absence of a rigidly structured caste system, people may retain strong caste attitudes and practices, creating what Wacquant (2024, p. 268) refers to as a general ‘“race custom” on the streets’ that maintains hierarchies of division and social exclusion. White workers’ attempts to instil moral boundaries in racially diversifying work forces, in Johnson’s (1936, 1941a) analysis, were not state-sponsored, but rather derived from caste beliefs and the desire to create a caste system in the economy at the moment where caste itself seemed to be disintegrating. Indeed, this is not too different from the aforementioned argument made by Davis et al. (2022[1941]) in Deep South, when they argue that there is much stronger emphasis and symbolic value placed on deferential behaviour from Black to white people in the Old City, as white people try to solidify the caste system through their beliefs and practices in a context where the economic relations seem to be liquidating the caste system.
As highlighted in the work of Johnson, therefore, we again see that classical race-caste authors were not brazenly equating race and caste. Without equating race and caste, Johnson showed how there can be ‘lags’ between an economic structure of society, which is decreasing in its caste-like nature, with the caste-like beliefs of the members within that same society. Here, therefore, we again see that the race-caste approach does not equate through comparison, but rather orients itself towards understanding a key general sociological problem: the relationship between ideas, practices, and social structure. Furthermore, despite Cox’s (1942) insistence that the race-caste approach has no hypothesis/es, we again see how classical authors precisely were generating hypotheses from thinking about these moments of habitus clivé. Johnson and Frazier, for instance, both looked at those moments of dissonance to hypothesise several claims that could be tested by empirical research, such as that capitalist social arrangements would lead to the dissolution of caste beliefs, or that caste beliefs become strengthened when caste status seems to be liquidating.
Concluding thoughts: from ‘reckless conflation’ to ‘scientific profits’
While the race-caste debate may seem to have disappeared into the annals of sociology (Yengde, 2024), this classical debate is being revisited in certain circles. Just like the classical era, this debate carries a degree of acrimony. Just as Pal (2025, p. 63) is arguing that with regards to race and caste ‘their reckless conflation [. . .] is an affront to sociological rigor’, others, such as Wacquant (2024, p. 274), insist that the race-caste approach provides ‘scientific profits [. . .] conceptual construction, clarification, and consolidation’.
In this paper, I have avoided saying whether we ought to understand race relations as caste relations. Once we acquaint ourselves with the classical literature on the race-caste approach, the question as to whether we ought to understand race relations as caste relations appears to be a question to be answered through empirical research rather than a conceptual assumption that can be made a priori. Classical advocates of race-caste approach, such as Davis, Drake, Johnson, and others, clearly argued that race relations can become caste-like, and race relations can also not be caste-like – they were by no means committed to the view that all race relations are caste-like relations.
The point of this paper, therefore, was to stick closely to the classical texts of the race-caste approach. In doing so, I have shown how critics of the race-caste approach fundamentally misrepresent these classical texts, often conducting multiple methodological errors in their criticisms. In contrast to these criticisms, I provided an expanded account of the work of classical authors of the race-caste approach. Here, I have shown that advocates of the race-caste approach did not see race relations as being directly equivalent to caste relations; rather, such advocates saw race relations as, in historically and geographically specific moments, becoming ‘caste-like’, where different levels of society (e.g. the economic, political, sexual, social, educational etc) can also be organised at different points along this caste-continuum. Put uncontroversially, all these authors were arguing was that in specific moments, race relations can be organised in a rigid way that becomes caste-like.
Such authors who used the race-caste approach argued that there were clear scientific benefits to their position. On the one hand, they argued this approach enabled them to generate a comparative sociology of hierarchy, that was able to compare: race relations across different local (e.g. rural vs urban), national (e.g. US South vs North) and transnational (e.g. US vs Kenya, South Africa vs Kenya) contexts, racial and non-racial hierarchical relations transnationally (e.g. race in the US South vs caste organisation in South Asia), and relations across time (e.g. pre and post-industrial periods of the US South). Through such comparative research, the race-caste scholars thus saw themselves as producing conceptually innovative, hypothesis driven sociological research. Furthermore, rather than assuming universally that all race relations were caste-like, classical authors of this tradition were keen to emphasise the historical specificity of their claims.
In sum, even if one disagrees with their analysis, it is improper to dismiss the work of classical authors of the race-caste approach ‘as an affront to sociological rigor’ (Pal, 2025: 63). The race-caste scholars of the classical era clearly developed a sociological programme oriented not towards demonstrating the substantive equivalence between race and caste, but rather oriented towards the relational empirical question of when, how, and why, in certain specific moments, racial domination becomes caste-like. In a context where the scholarship of scholars such as Frazier, Johnson, Davis, and Drake tends to be excluded from the sociological mainstream (Meghji, 2024, 2026), we would do well to note their scholarship on race-caste as a crucial space of innovation. Through developing this race-caste approach, this group of scholars refined both sociological methods (e.g. comparative ethnography, and comparative-historical sociology), while they also tackled some of sociology’s ongoing riddles – such as the relationship between beliefs, attitudes, practices, and social structure.
Of course, readers of this paper may be curious as to my own position on the race-caste debate, and whether I think we ought to revive this approach in the 21st century. My answer is that as social scientists, rather than philosophers, the proof of the pudding lies in the empirics; classical scholars did not waste time debating ‘is this race or caste?’, but rather used the notion of a caste continuum to guide (often comparative) empirical research into the relative rigidity of a given structure of racial domination. As to whether or not we should revisit this approach in contemporary research, I will simply repeat the point that we can only answer this through empirical testing.
Footnotes
Funding
The author disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Funding provided by the British Academy, grant no. SRG2324\241031.
Declaration of Conflicting Interests
The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.