Gentlemen,

let me answer the question "What is a representative sample?" as follows.

A representative sample have to reflect as the intrinsic properties of the object to be evaluated as the requirements of the user (the cost, as an example)to estimate these properties.

A repsentative sample has to represent the material th be evaluated under given economic and other restrictions.

That is,

- the sample has to be precisely (the variance of the measured value must not exceeed a given upper bound)

- the sample has to be unbiased (the true but unknown content is estimated by the mean value without any systematic error)

- the sample has to be reproducible, i.e., by repeating the sampling under one and the same conditions, statistically identical results are obtained

and finally

- the sample must be reliably, i.e., a certain suitable defined risk (cost, variance of the mean and others) must not exceed a given upper bound or must take a minimum.

Obviously, these properties of a representative sample depend on the strategy of sampling as well as on the number of individual sample taken.

By the way, the representativeness of sampling and instrumental measurement will be discussed on the 8-th international "Freiberg sampling symposium". The symposium is organized by the IQS Freiberg and will be held on 8-th to 9-th november 2002 in Freiberg.

You are invited to participate and to give a lecture.

Announcements are received via e-mail: "rasemann@geo.tu-freiberg.de"

Regards

W.Rasemann ■