Tomlinson, R. C.
Member of Staff, Scientific Department, National Coal Board, London,
Pages: 22 Published: Jan 1955
Interest in coal sampling in the United Kingdom has received a great impetus in recent years for a number of reasons. The nationalization of the coal industry itself and of two of the major coal-consuming industries, gas and electricity, has resulted in three very large commercial organizations with a large annual sampling bill. Then, more recently, the National Coal Board introduced a revised price structure whereby the pithead price of a coal is determined largely by its chemical analysis. This has meant that, more than ever before, there is a commercial incentive towards obtaining detailed information on the properties of the many hundreds of grades marketed each day. This increased commercial interest has arisen at a time when the introduction of various new ideas about sampling had led to a renewed theoretical and statistical examination of the subject. As a result much experimental work has been carried out, providing a considerable amount of sampling data. The results of some extensive experiments carried out for the British Standards Institution in 1950–1951 have been reported by E. H. M. Badger, Chairman of the committee concerned (1). The main conclusions, so far as they affect a sampling specification, were that the weights of sample prescribed in the existing standard were not far from correct but that two changes would be required in future specifications: first, more attention would have to be paid to the errors of reduction and analysis, particularly when an attempt is made to obtain the very high degree of accuracy demanded by engineers in charge of boiler trials; second, provision should be made for the measurement of sampling accuracy from the results of routine sampling using the technique known as duplicate sampling (2). A further conclusion, and one of perhaps wider importance, was that the science of coal sampling could not be built on theories of coal sampling but must be based empirically on measurements carried out under a wide range of conditions. This is no new discovery, but until recently there was no very satisfactory technique for carrying out the measurement. However, with the help of some of the techniques used by statisticians for other sampling problems, the work of measurement is now moving forward, and indeed the main theoretical interest has been in developing further new and simple techniques. It is the intent of this paper to describe the line of effort which has been carried out during the past three years, and to illustrate those techniques of measurement which seem to be of most interest and lasting value, and finally to describe the way in which the conclusions that have been reached are being incorporated into a revised specification. Much remains to be done, however, before the revision of this specification can be completed.
Paper ID: STP46263S