19:3.] STEVENSON— FORMATION OF COAL BEDS. 103 



upper and lower benches of 16, 13, 14, 12, 11, 9, 8, 7, 5, 4, 3, and 2 

 per cent. Sometimes the upper, at others, the lower is the less clean. 

 Often there is no difference in appearance but usually the cause of 

 greater impurity is distinct, for the filmy partings of black shale are 

 apparent. Local variations are equally well marked in analyses by 

 the same chemist. The ash in the Lower Freeport varies from 1.80 

 to 10.53 per cent, and in the Middle Kittanning from 3.48 to 12 per 

 cent., the samples in every case being of coal which is mined. It 

 must be evident that a collection of analyses from all regions cannot 

 be utilized for generalization. Conditions varied locally at each 

 horizon, so that while worthless coal was accumulating in some 

 places, good coal was accumulating in others ; equally, the condi- 

 tions varied greatly during the period of formation, so that one 

 bench may be clean and another worthless. 



But a promiscuous collection of analyses is not merely worthless 

 as a basis for generalization, it is also very apt to be misleading by 

 diverting one's attention from consideration of the features which 

 are really important. One is not concerned with averages of coals 

 all over the world or in the proportion of analyses showing more or 

 less than 5 per cent. The really important matter is the composition 

 of a particular deposit within a large area. When this has been 

 ascertained, one finds that the difficult problem is not to account for 

 the excess of ash but for the astonishing deficiency in ash, observed 

 in some beds throughout very great areas. Analyses by A. S. 

 McCreath and by Hite and Patton^"^ show that the Campbell's Creek 

 coal bed in 4 counties of southern West Virginia gave as the result 

 of 26 commercial samples, 5.943 per cent, of ash and all were of 

 outcrop coal, yet 10 of them had less than 5 per cent. In 4 other 

 counties, the average of 34 commercial samples is 5.52 per cent, and 

 several had less than 3. Commercial samples of the Pocahontas 

 coal from 38 localities in southern West Virginia showed from 2.34 

 to 9.58 per cent., with an average of 4.63. The Pittsburgh usually 

 has 7 per cent, or less. All of these are in areas of from 2,000 to 

 6,000 square miles or more. Evidently there are coal beds which in 



"'W. Va. Geol. Surv., Vol. II., 1903, Pp. 695, 696; Vol. II. a, 1908, pp. 

 393, 394- 



