144 ANTHROPOLOGICAL FORUM for example, mana. Thus, the term'supernatural'can be used to describe beliefs relating to spirits Furthermore, it is impossible to 'discover' the correct definition of any concept; we label things, and give a definition to our concepts, indicating how we intend to use them. Chinese people often claim to have no religion, but anthropologists have not hesitated in defining the complex of activities including ancestor worship, worship of deities in the home, and visits to temples as popular religion. Contemporary Chinese usage defines religion as institu- tionalised religions such as Buddhism or Christianity. The term 'religion zongjiao, is, in fact, a modern term combining two characters, zong(ancestors and'to venerate,) and jiao(,). Most students in Hong Kong claim they have no religion, though they and their families participate, in varying degrees, in popular religion. In Taiwan, informants, especially college-educated ones, claimed that ancestor worship and lunar new-year rituals were not religious, but just 'customs. We need not use their definition of religion, however. There have been debates in the anthropology of China about whether to view popula religion as a single religion or many (e. 8, see Bell 1989; Freedman 1974; Wolf 1974), but I know of no anthropologist who accepts the popular notion that customs are not part of religion. It is not ethnocentric to use our anthropological concept of religion rather than my informants'concept, since it allows us to ompare aspects of Chinese popular religion with other religions. We may also jettison the concept of religion and usecosmologies' instead. In any as we can use our anthropological definition of religion or cosmology, we can use an anthropological definition of supernatur Klass (1995: 28-30) illustrates his argument that it is ethnocentric for the anthropologist to impose his or her view of reality by citing the case of Trinidadian farmers for whom rent to an absentee landlord and offerings to the di, a spirit of the field, are equally real. Klass overstates his case when he claims that, were a young leftist to believe that landlords had no right to the crops, the andlords would not exist. This is an error of logic; even if for the young leftist landlords should not exist, he would do well to realise that they do exist. Presumably, they or their agents would come to claim the rent if it were not paid Chinese traditionally also have had earth gods and deities comparable to the di. Because these spirits do not actually consume the food left for them (in contrast to landlords who actually did take their rent, which was once about half the crop), Chinese informants recognise the spirit world as different from the human world. Chinese people see the heavenly, human, and underworld ian,ren,di)domains as linked, but they recognise them as different. Indeed, it is precisely because they are different that rituals and offerings are necessary to link humans and spirits. Thus, though Klass argues that paying rent to the landlord and making offerings to deities of the field could be viewed as the same by the farmer i wonder if they are not actually viewed differentl144 ANTHROPOLOGICAL FORUM for example, mana. Thus, the term ‘supernatural’ can be used to describe beliefs relating to spirits. Furthermore, it is impossible to ‘discover’ the correct definition of any concept; we label things, and give a definition to our concepts, indicating how we intend to use them. Chinese people often claim to have no religion, but anthropologists have not hesitated in defining the complex of activities including ancestor worship, worship of deities in the home, and visits to temples as popular religion. Contemporary Chinese usage defines religion as institu￾tionalised religions such as Buddhism or Christianity. The term ‘religion’, zongjiao, is, in fact, a modern term combining two characters, zong (‘ancestors’ and ‘to venerate’) and jiao (‘teachings’). Most students in Hong Kong claim they have no religion, though they and their families participate, in varying degrees, in popular religion. In Taiwan, informants, especially college-educated ones, claimed that ancestor worship and lunar new-year rituals were not religious, but just ‘customs’. We need not use their definition of religion, however. There have been debates in the anthropology of China about whether to view popular religion as a single religion or many (e.g., see Bell 1989; Freedman 1974; Wolf 1974), but I know of no anthropologist who accepts the popular notion that ‘customs’ are not part of religion. It is not ethnocentric to use our anthropological concept of religion rather than my informants’ concept, since it allows us to compare aspects of Chinese popular religion with other religions. We may also jettison the concept of religion and use ‘cosmologies’ instead. In any case, just as we can use our anthropological definition of religion or cosmology, we can use an anthropological definition of supernatural. Klass (1995:28–30) illustrates his argument that it is ethnocentric for the anthropologist to impose his or her view of reality by citing the case of Trinidadian farmers for whom rent to an absentee landlord and offerings to the di, a spirit of the field, are equally real. Klass overstates his case when he claims that, were a young leftist to believe that landlords had no right to the crops, the landlords would not exist. This is an error of logic; even if for the young leftist landlords should not exist, he would do well to realise that they do exist. Presumably, they or their agents would come to claim the rent if it were not paid. Chinese traditionally also have had earth gods and deities comparable to the di. Because these spirits do not actually consume the food left for them (in contrast to landlords who actually did take their rent, which was once about half the crop), Chinese informants recognise the spirit world as different from the human world. Chinese people see the heavenly, human, and underworld (tian, ren, di) domains as linked, but they recognise them as different. Indeed, it is precisely because they are different that rituals and offerings are necessary to link humans and spirits. Thus, though Klass argues that paying rent to the landlord and making offerings to deities of the field could be viewed as the same by the farmer, I wonder if they are not actually viewed differently