A clarification of the concept of dependent selection of particles

A generalization of Gy?s 1979 model for the variance of the fundamental sampling error has been proposed at WCSB2?2005 by Geelhoed. This generalized model can potentially lead to a better correspondence with actual sampling processes. However, a new parameter is required by the new model: ?the parameter for the dependent selection of particles?(Cij). It is important that this new parameter, which is at the basis of the new theory, is well understood in the sampling and blending community. In order to clarify the meaning of Cijand to prevent potential misinterpretation of Cij, we discuss the difference between Cij and an intuitively associated, but different, parameter: ?the parameter for the dependent selection of particle types?(Mij). In this contribution, the theoretical difference between both parameters is clarified. Simulations demonstrate that, numerically, the difference between Cij and Mij can be large. Some additional advantages of the approach which uses Cij are discussed: (i) the approach based on Cijis more general; (ii) in all cases where Mij is known, Cij can be determined as well, while the reverse is not true; (iii) an equation already exists that relates the sample variance to the value of Cij.