clarkk.] AVERAGE COMPOSITION OF ROCKS. 17 
To the -statement of a "trace" I have arbitrarily assigned the value of 
0.01 per cent. In this way I have obtained the following averages, 
which represent all of the data accumulated by the chemist of the 
Survey down to January 2, 1904. 
Silica. — Average of 1,358 determinations, 60.91 per cent. 
Alumina. — Average of 912 determinations, 15.28 per cent. In 93 additional deter- 
minations, Ti0 2 and P 2 5 were not separated; and these, therefore, are not 
included in the computation. 
Ferric oxide. — 961 determinations. Average, 2.63 per cent. 
Ferrous oxide. — 962 determinations. Average, 3.46 per cent. In 38 analyses the 
oxides of iron were not separately determined, and in them the figure for iron is 
excluded from the calculation. 
Magnesia. — 1,027 determinations. Average, 4.13 per cent. 
Lime. — 1,215 determinations. Average, 4.88 per cent. 
Soda. — 1,268 determinations. Average, 3.45 per cent. 
Potash. — 1,265 determinations. Average, 2.98 per cent. 
Water. — In 288 analyses, total water was determined; average, 1.86 percent. In 626 
analyses, water lost below 100°-110° was estimated; average, 0.44 per cent. In 
688 analyses, water retained above 100°-1 10° gave an average of 1.51 percent. 
Combining these data, the probable percentages are, 11,0 — , 0.41, and H 2 0+, 
1.49 per cent. 
Titanic oxide. — 870 determinations. Average, 0.73 per cent. 
Zirconia. — 185 determinations. Average, 0.033 per cent. These determinations were 
not made so generally as to give the average any great importance. 
Phosphoric oxide. — 884 determinations. Average, 0.26 per cent. 
Baryta. — 617 determinations. Average, 0.11 per cent, 
Strontia. — 520 determinations. Average, 0.04 per cent. 
Manganese cm'de.— 899 determinations. Average, 0.10 per cent. 
Carbon dioxide. — 169 determinations. Average, 0.53 per cent. This is doubtless too 
high, for in many rocks carbon dioxide was not looked for. In 142 of the 
analyses in which it was reported the percentage was zero. 
Lithia. — Commonly reported in traces, but often absent. Reckoning a "trace" as 
0.01, the average of 550 analyses is 0.011 per cent. This is probably not far 
from a true estimate. 
Nickel oxide. — 243 determinations. Average, 0.029 per cent, which is probably too 
high. 
Chromic oxide. — 246 determinations. Average, 0.053 per cent. Probably too high. 
Vanadium trioxide. — Only 40 determinations are reported; in average, 0.026 per cent. 
Determinations too few to give this value any weight. The figure, however, is 
not without interest. 
Chlorine. — Determined or proved to be absent in 234 cases. Average, 0.07 per cent. 
Probably too high. 
Fluorine. — Only determined or proved to be absent in 73 analyses. Average, 0.12 
per cent, which is undoubtedly too high. An inferior limit, however, may be 
fixed with reference to P 2 5 . If we assume that to represent apatite the equiva- 
lent amount of fluorine should be 0.023 per cent. The true average must be 
slightly higher. 
Sulphur. — This element is reported in three forms — as S, as S0 3 , and as FeS 2 . 
Reported as S, 188 analyses give an average 0.045 per cent. As S0 3 , 259 analyses 
give 0.11 per cent, equivalent to 0.044 per cent of S. In 128 analyses 0.60 per 
cent FeS 2 was found, equivalent to 0.32 of sulphur and 0.28 of Fe. The latter 
must be taken into account in estimating total iron. General average for sulphur, 
0.108 per cent. 
Bull. 228—04 2 
