The logarithmic series is sometimes inadequate as a descr. of sp.-frequency in samples from natural populations. In these cases, the use of more general, less simplified, negative binomial series, adjusted for removal of the zero term, should be considered as an alternative to the truncated log-normal function proposed by Preston. These two differ fundamentally in the form of population assumed. Whereas latter supposes that rare spp. are few, the former, although allowing for this eventuality, has so far only been found to exist in the form where rare spp. are the most numerous.