In the tarantism cult of southern Italy there are two types of actual tarantula spider with which the spirit spider might be identified, but only one is selected as spirit counterpart. In examining the basis of symbol selection here, reference is also made to Lele pangolin symbolism where there are again actually two species of local pangolin, only one of which is paid cult. Mary Douglas's classic analysis ofanimal 'anomalies' fails to address this problem. In the context ofa comparison ofpangolin ritualisation in other Congolese cultures, the following critique of Douglas's treatment of pangolin 'anomalies' suggests a cosmological solution to this puzzle. Douglas's attempt to correlate pangolin power with exogamy is also shown to be questionable. More generally, it is argued that where two or more related animals are subject to differential symbolisation a satisfactory interpretation should explain why one is selected for preferential ritual attention. I In her seminal essay on Lele religious symbolism, Mary Douglas (1957) rightly observed that, in order to be effective, symbols do not need to be grounded in objectively accurate knowledge and may indeed be based on mistaken information. So, in the case of animals selected to act as symbols, what matters is not actual animal behaviour but rather the characteristics and properties imputed to them by those who employ them as symbols. The famous tarantismo spider cult of southem Italy shows how this dictum may sometimes be applicable at one level while not applying at others. The tarantula spirit spider in Apulia (the region containing the eponymous town of Taranto) has potentially two types of actual tarantula spider (Latrodectus and Lycosa) with which it might be identified, but only one (Lycosa) is selected as the actual counterpart for the spirit. The spider (Latrodectus) which is passed over is, paradoxically, armed with a toxic bite capable of indeed causing the symptoms associated with tarantism. This situation of two similar zoological categories, only one of which receives ritual attention, has parallels in Lele animal symbolism (and is not unknown elsewhere). Douglas bases her famous analysis of the 'anomalous' pangolin on one type (Manis tricuspis) and rather passes over the other pangolin (Manis gigantea) which is also found among the Lele. As we shall see, this raises questions about the adequacy of her anomaly argument and invites further consideration of her data. More generally, where such pairs of related zoological avatars are subject to such differential symbolisation this offers a unique opportunity to test the validity of theories of symbolisation. A fully satisfactory interpretation should be able to explain the circumstances in which one of the pair is chosen at the expense of the other.