Cu, Pb, Zn And Ba In Carbonate Residuum Of Southern Missouri

The trace element geochemistry of the cherty, clay-rich carbonate residuum. Covering much of southern Missouri and adjacent parts of Arkansas was investigated during the early 1970's as part of a statewide geochemical survey of Missouri-1/2/3 The study was environmentally oriented and focused on the character of the "natural' geochemical environment of the state. A great deal of effort was made to collect samples that would represent the geochemical variability of the near-surface environment in an unbiased way. In general, this was accomplished by use of randomization procedures in both collection and analysis. Most of the sampling sites were selected randomly to prevent personal bias from entering into sample selection. In the laboratory, samples were analyzed in a randomized sequence to circumvent the effects of any systematic bias that may have entered into the laboratory procedures. Several residuum samples contained 1 000 parts per million (ppm) or more of Pb, Zn and Ba. All these samples were collected in Washington County in or near the barite district. The frequency distributions shown in figure 1 indicate just how unusual the residuum in Washington County is with regard to these three metals and copper. The shaded parts of the histograms represent the samples from Washington County. The histograms are based on analyses of 141 samples from 38 localities scattered over the subcrop of the Cambrian and Ordovician Formations of southern Missouri and northern Arkansas. Eight localities were selected randomly from over each of four bedrock units--Bonneterre Dolomite, Potosi and Eminence Dolomites, Roubidoux Formation, and the Jefferson City, Powell and Cotter Dolomites. Six localities were selected randomly from over the Gasconade Dolomite. The geochemical "background" for these elements in residuum is defined by samples from outside of Washington County (unshaded part of the histogram); lognormal distribution curves have been fitted to the corresponding parts of each histogram. Clearly, these lognormal models (curves) describe well the background