Reference samples in geology and geochemistry

F.J. Flanagan

United States Department of the Interior, Geological Survey,
Reston VA. 22092, USA.

Abstract
The use of geologic reference samples may have started in the last quarter of the 19 th century after analysts in England, Germany, and the United States deplored the large variation in data obtained by different analysts for the same constituents in the same sample. Early samples involved in cooperative analysis, also known as "round robins", were mainly industrial or agricultural materials of economic value, or products derived from them. Data from some round robins were so diverse that "uniformity in analysis" was a part of early committee names. The first two geologic samples of the National Bureau of Standards were analyzed cooperatively. Samples for geology and geochemistry are classified as artificial samples or as natural materials of either economic or geochemical interest - the latter distinction not always possible. Artificial samples have been prepared in several substrates, and samples in other substrates may be limited only by the ingenuity of the analyst. The principal suppliers of samples of economic interest are discussed, by country, to show the range of samples prepared by organizations. Natural samples of geochemical interest are mentioned by sample type. Because of their intended use, some geochemical samples are restricted to analysts using specific techniques. Methods used for and precautions to be observed while preparing large or small rock samples are discussed. Excess grinding of small rock samples reduces the amount of FeO found in a rock analysis. Unavoidable contamination occurs when processing rocks in steel. Tungsten carbide, and some mechanical equipment , should be avoided. A reviewed of the allied subjects of homogeneity of a sample, sampling errors, and size of the sample grains shows that we have reinvented the wheel since 1885. Methods used to (1) test the homogeneity of samples, (2) eliminate outliers which some scientists reject repeatedly despite warnings of statisticians, and (3) estimate best values when the method for estimating should be part of the program for a sample are discussed. Calibration lines may be used to determined if "best values" in a series of several samples are really "best and to furnish confidence limits for a single datum. The agreement of data by an analyst for a reference sample with data preferred by an issuing organization is usually described by one of several adjectives, a subjective procedure. A variation of a c2 method may be used if a producer specifies the exact number of data for the cooperative analytical program to be furnished by each analyst and then calculates the population estimates of the mean and standard deviation for an element in a sample. Because of the large increase in the number of reference samples of all types, a revised table listing samples that may be useful in geochemistry is included. It is known is the samples, other than those of the U.S. Geological Survey, were collected in conformance with provisions of the Plant Pest Control Regulations of the U.S. Department of Agriculture, or of similar organizations in other nations.

Geostandards Newsletter (1986), Vol. 10 No. 2 pp. 191-264

Copyright   © Association Scientifique pour la Géologie et ses Applications
Vandoeuvre-lès-Nancy, France.