Quantitative determinations were made of the banded ingredient composition of a number of domestic stoker coals prepared for combustion tests from large lot samples of small-size coal procured at 15 Illinois coal mines.The determinations were made by identifying and counting the discrete coal particles in selected samples using the petrographic technique which is commonly used to assay performance in ore dressing operations.A standard procedure for the analysis was developed which included close control of all steps in selecting the sample, crushing to free coal components into discrete particles, and size classifying the particles by mechanical screening. Strict criteria were adopted for identifying the coal particles.The results of the petrographic analyses which were of special interest in connection with the combustion tests concerned the effect of coal cleaning processes upon the proportion of banded ingredients. An increase in vitrain content of the cleaned coal was usually noted but concentration of this ingredient was never of considerable amount. The analyses also showed there was no great concentration of vitrain in the lots of small-size coal secured from the mines. Previously it had been thought that vitrain by virtue of its friable nature constituted a major proportion of the "screenings" produced in coal mining and preparation.During the course of the work certain incidental investigations were carried on concerning the effect of differences in hardness of the banded ingredients on size distribution when more or less "pure" ingredients were crushed under controlled conditions. The variation in specific gravity of the banded ingredients and the effect of this variable property in separating or concentrating coal ingredients by use of heavy liquid media was investigated. The paper discusses reliability of this method of petrographic analysis of coal and reports certain experiments made to determine degree of accuracy achieved.