162 



SCIENCE 



N. S. Vol. XXX. No. 762 



work was not limited to analyses alone; 

 problems for research, were frequently sug- 

 gested ; and a variety of interesting investi- 

 gations have been carried out in the survey 

 laboratories. My aim in this communica- 

 tion is to give a brief summary of the 

 results so far obtained, and to indicate 

 something as to their significance. 



In the first place, more than one thou- 

 sand analyses of igneous rocks have been 

 made, and also many partial analyses for 

 special purposes. This work was called for 

 by the petrographers of the survey, and it 

 has led to some unforeseen consequences. 

 In the beginning, the analyses were made 

 along conventional lines and the minor con- 

 stituents of the rocks were neglected, a 

 policy which is still followed in many Euro- 

 pean laboratories. It was soon found, how- 

 ever, that greater refinement of the work 

 was desirable, and that in some eases a rock 

 analysis, to be really satisfactory, needed 

 from fifteen to twenty separate determina- 

 tions. At present the following substances 

 are commonly determined, or at least their 

 absence is proved: silica, alumina, both 

 oxides of iron, magnesia, lime, soda, potash, 

 water at and above 100°, titanic oxide, zir- 

 conia, carbonic acid, phosphoric acid, sul- 

 phuric acid, sulphur, baryta, strontia, 

 manganese, nickel and chromium; and 

 often also vanadium, chlorine and fluorine. 

 Lithia is sometimes determined, but pos- 

 sible boron and glucina are ignored. The 

 water is always determined directly, never 

 by loss on ignition; and recently, because 

 of the researches of Mauzelius, as verified 

 and extended by Hillebrand, the influence 

 of the fineness or coarseness of the rock 

 powder is taken into account. We now 

 know that in fine grinding, part of the fer- 

 rous oxide in a rock or mineral becomes 

 oxidized, and also that notable quantities 

 of water are at the same time absorbed 

 from the atmosphere. The errors thus in- 



troduced into an analysis are too large to 

 be neglected. As far as possible the an- 

 alytical operations are performed in plati- 

 num dishes, in order to avoid contamina- 

 tions derived from glass. Furthermore, all 

 reagents are carefully tested as to their 

 purity. Without these precautions accu- 

 rate work can hardly be done. 



The analyses thus made are evidently 

 much more complete than those collected in 

 Roth's well-known tables, and at the same 

 time their summation is better. To be 

 satisfactory a rock analysis should sum up 

 between 99.50 and 100.50 per cent.; the 

 average summation error of the survey an- 

 alyses being only 0.20 per cent. The de- 

 terminations of the minor constitutents not 

 only render the analyses more precise, but 

 also aid the petrographer in his discussion 

 of the data. For example, the average 

 proportion of titanic oxide found in 989 

 analyses is 0.74, and that of phosphoric 

 oxide is 0.26 per cent., or one per cent, 

 together. If these substances are not de- 

 termined they render the determination of 

 the alumina too high, and when the petrog- 

 rapher attempts to compute the proportion 

 of feldspar in a rock the error is multiplied 

 several times over. The slight value of 

 analyses in which these determinations are 

 neglected is therefore almost self-evident. 



For the work thus done upon the igneous- 

 rocks, chief credit must be given to Dr. 

 Hillebrand. He first perceived the impor- 

 tance of more thorough analyses, and did 

 much toward perfecting the analytical 

 methods. His bulletin, "On the Analysis 

 of Silicate and Carbonate Rocks," is al- 

 ready recognized as a classic.^ 



From a small number of rock analyses 

 only minor conclusions can be drawn. 

 They aid the petrographer in the study of 

 his special group of specimens ; but, unless 

 they are correlated with other data, they 



= Bulletin 305, U. S. Geological Survey. 



