It has often been suggested, and it is more or less commonly believed, from the consideration of the density and the magnetic character of the earth and from the knowledge of the composition of many meteorites, that part, at least, of the earth's interior is composed largely of iron or of a nickel-iron alloy similar to that which constitutes the iron meteorites.
1. DEPARTMENT OF THE INTERIOR HUBERT WORK, Secretary UNITED STATES GEOLOGICAL SURVEY GEORGE OTis SMITH, Director Professional Paper 127 THE COMPOSITION OF- THE EARTH'S CRUST BY FRANK WIGGLESWORTli CLARKE AND HENRY . STEPHENS WASHINGTON . WASHINGTON. GOVERNMENT PRINTING OFFICE 1924
2. ADDITIONAL COPIES OF THIS PUBLICATION MAY BE PROCURED FROM THE SUPERINTENDENT OF DOCUMENTS GOVERNMENT PRINTING OFFICE WASHINGTON, D. C. AT 20 CENTS PER COPY .
3. CONTENTS. Pag~. Introduction ............................................................ -................_........ ·....... . ] The crust and the interior .......................................................................... . 1 General features of igneous rocks ............ ·.............. ·.......................................... . 2 Mineral constituents pf igneous rocks ............. : .......... ~ ........................................ . 3 Chemical constituents of igneous rocks ....................................................... .- ....... . 4 Average composition of igneous rocks ......................................... : ..................... ·...... . 6 A.verages con1pu ted ..................................................................... : . ......... . 6 .Methods of averaging ............ ~ .................................................................. . 10 Areal averages ................................. : ... ~ ... , ...... ~ ..................................... . 12 1'he general average ........................ , .................................... _. .................... . 16 Relative abundance of the elements in the earth's crust .................................................... . 1.6 Masses of the lithosphere, the hydrosphere, and the atmosphere ............. ~ .......·.................... . 1.6 Mass of the lithosphere .......................................................................... . 16 .Mass of the hydrosphere .....·.............................................................. : .... . 17 Mass of the atmosphere .. ·: ................................. ·....................................... . 17 Composition of the crustal mass ..........·... : . ... ·.................. ·.......... ·........... ·......... . 17 Average elementary composition of igneous rocks ...... .'............................................... . 1.9 G·eneral staten1ent ................................................................................ . 19 ])ata considered ........................ ·: ......... : ...................... : ....................... . 21 Average composition of sedimentary rocks ........................................... ·................. . 28 Average composition of the lithosphere ........................ , .... ~ ............ ·.. ·................... ·.. 30 Average composition of the hydrosphere ............................................................. . 33 Average composition of the atmosphere ............................... ·......... ; ...................... . 33 Average composition of the earth's crust ............................................................. . 33 Average composition of stony meteorites ..................... , ...... ·...... ·............................ . 34 Accuracy of the analyses ........... : .................................................................... . 35 Questions to be considered ......... ~ .............................·.. ~ ......................... : ..... - .. 35 Probable errors in the determination of the major constituents ......................................... . 36 Silica ......................................... , ................................................. . 36 Alumina .......... ·................................·....................... : ..................... . 37 Ferric oxide ........................................... ~ ................................... ; .... . 37 li'errous oxide~ .......... ·...... :.' .................................. " .............................. . 37 Lime: ................................................................................. · ~ · , · ~ · · ·. · 38 Magnesia ............................ .- ..................................... ·..... ·......... : ..... . 38 Soda and potash: ...................................... : ........................................ . 38 \Vater .................... ~ ........... ~ ......................................................... . 38 ~eitanium dioxide .......................................................................... ·.... . 38 Phosphorus pentoxide .. ·........................................................................ . 39 Manganese oxide ................................... ,; ................ :; ....... ·, ............. -':: .... . 39 Probable errors in the determination of the minor constituents ..................................... : ... . 39 Zirconia ................................................................... ,. ..................... . 39 ltare earths ............................. : ...................................................... . 40 Sulphur ........ : ..............................................................· .............. ·.... . 40 Chlorine ..............·..................... ·.................. ·............. -.. ·.. :.:.:.~.· . ." ......... . 40 Fluorine .......................................................................... ·............. . 40 ·. ·. ·. ·. ·. ·. ·. ~:~~~: :::~~~~::!e~. ~: ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ -~:~ ~: ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~·~ ~ ~ ~ ~ ~-~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ -~ ~ ~ ~· ~ ~ ~ ~ ~ 40 40 ~~~~a~l-~~~~~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~-~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~: ~ ~-~ ~-~-~ -~ ~ ~>_ ~ ~ ~ ~. ·~ ~~ ·40 ... ·. ·. : ·: : :. : : : : :::: : : :: : : : : ·: ·.· ." :: ·. ·. ·. ·: ·. ·. ·. ·. ·. ·. 41 Strontia ........................................................................................ . 41. Lithia ........................... : ............................................ : ................. . 41 Copper ..................... .' ................................ ; ... " ........................ : . ..... . 4.1 Geographic r~sum~ and characterization of the data ............ ·.......................................... . 4l ~~~: ~!:;:::~~~~~~~---.·.·--~:: ~ ~-~ ~ ~ ~ ~ ~ ~-~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~-~ ~ ~ ~ ~ ~ ~-~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~::: 41 41 Greenland ...................................................................................... . 41 East Canada ................................................................................... .- . 42 Alaska .................................... : ......... : .................... ·.··.~··.············; ...... ·. 43 III
5. CONTENTS. v A:.,;soeiation and distt'ibution of the elements-Continued. Helati.ons of the petrogenic elements-Continued. Page. Nitr.ides and nitrates ............................................................................ . 85 Phosphides and ph,osphates .............................................................. ~ ... : . .. . 86 Uolumbates and ta.ntaJates .................................................... , .................. . 86 Chroma.tes and molybdates ...................................................................... . 87 Sulphides ..................................................................................... . 87 Fluorides ............. _. ...... : ............................... ·................................... . 88 Chlorides ........................................................... ·........................... . 88 A.rsenides to iodides ..................................................... ·........ ·............... . 88 Relations of the metallogenic elements ............................................................... . 89 Native elen1ents ............................. ·.................................................. . 89 Oxides 89 Silicate~~:.·.·.·.~~::::::::::::::::::::::::::::::::::::::::::::::·:::::::::::::::::::::::::::::::~~~: 90 Chromates and molybdates ................................ ·." ..................................... . 90 Arsenides, antimonides, and bismuthides ........................................................ . 90 Sulphides ..... ~ ............................................................................... . 91 Selenides a.nd tellurides ..... : ................................................................... . 91 Sulpho-salts .............................................................. : . ............ ·........ . 92 llalides ........................................................................................ . 93 Suuunary ........................................................................................... . 93 General relations .................................. ·.................... : ......................... . 93 Elements having doubtful relations ............................................ ~ .................. . 96 Differences between the two groups ............. : . .............................................. . 97 Spectroscopic differences ........................................................................ . 98 Elements found in meteorites .... : ............................................................... . 98 Correlation of the elements ..............................·............ : . .................................. . 99 Evolution of the elements ........................................ ·.....·........· ......................... . 107 Index ................................................................ ·..... : .......... ·................. . 113 ri'ABLES. Page. l. Previously published estimates of the avemg~ composition of igneous rocks....... . . . . . . . . . . . . . . . . . . . . . . . . 8 2. Average percentages of the minor constituents shown by 5,159 analyses ............. , . . . . . . . . . . . . . . . . . . . :10 3. Average composition of the igneous rocks of northern North America .....·................................ 13 4. Average composition of the igneous rocks of the United States. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 5. Average composition of the igneous rocks of Central and South Ame~ica ............................·..... 14 6. Average composition of the igneous rocks of Europe..................................................... 14 7. Average composition of the igneous rocks of Africa and Asia. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 8. Average composition of the igneous rocks of Australasia, Polynesia, and the Antarctic Continent ... ~.. . . . . . 15 9. Average composition of the igneous rocks of the continents and oceanic islands and of the earth............ 15 10. Average chemical composition of igneous rocks................... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 11. Average elementary composition of igneous rocks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 12. Average composition of sedimentary rocks ........................ .'. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 13. Average composition of the lithosphere ................................................................ :. 32 14. Composition of the oceanic salts ......................... ·.... :......................................... 33 15. Composition of the ocean ............................................. ·....................... ... . . . . . . . . 33 16. Volumetric composition of the atmosphere............................................................. 33 17. Elements in the lithosphere, the hydrosphere, and the atmosphe~·e .................... :. . . . . . . . . . . . . . . . . . 34 18. Average analyses of stony meteorites................................................................... 35 :UJ. Comparison of analyses used and .areas represented ............................. :........................ 70 20. Averages for the earth obtained by different methods ..... ~........................ ... . . . . . . . . . . . . . . . . . . 70 21. Periodic classification of the elements ... ~..... . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 22. Periodic arrangement of the elements ....... :.......................................................... 74 23. '!'he petrogenic and tjle metallogenic elements ........................................................ ~. 76 24. Groups of ininerals formed by the petrogenic elements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 25. Groups of minerals formed by· the metallogenic elements .. :.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 26. General occurrence of mineral groups iri. the ·earth....................................................... 97 27. H.elations of some oxides to silica and alumina ...................·. :. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 28. Periodic classification of the elements ......................... :........................................ 108
6. rfflE COMPOSI1'ION 01? r-rHE .E-AR1'H'S. CR·usr-r. By F. ,¥. CLARKE and I-I. S. WASHINGTON.· INTRODUCTION. 1 general are not notably 1nagnetic this· action THE CRUST AND" THE INTERIOR. may be attributed to the c}laracter or compo- The tern1 "crust of the earth" is a herit~ge sition of the n1aterial that forn1s the interior. fro1n the days when the interior of the earth 5. Frmn a study. of the propagation of the was generally conceived. to be a "sea of. n1olten shocks or waves of earthquakes, .we are led to roc}\:" n.t ·an enornwusly high tmnperature believe that there is a· cl1ange in the earth's covered with a relatively thin, solid crust of phy~icd properties at a depth of about 0.5 of cooled nuttter. V m·ious cogent reasons, which its radius, the materi.al below this depth not cn.n .not be cons:iderod here, have led to the t_rn.nsrnitting transverse vibrations. Studies abn.ndomnent of th:is concept, and we now of the ·cOJnpressibility of rocks and Jninerals have reason to hold the following tenets as to made by Adams and Williamson, 3 in the geo- the conditions that exist in ·the earth's interior: physical laboratory of the Carnegie Institu- 1. The interior n1ass is essentially or behaves tion, indicate that the high density of the essentia.lly like a rigid solid, although it may interior can not be· clue to con1pression alone, possibly have a certain degree of fluidity. so we nuist conclude that there is also· a change 2. The interior is probably hot, but it is of in the actual substance of the earth toward its unknown tmnperature, and its te1nperature center. 4 . pL'obably increases toward the center,. bQt 6. It has often been suggested, and ·it is through gradients that are undetern1ined be- 1nore or less conunonly believed, froni a con- yond very 1noderate depths and that are sideration · of the density and the 1nagnetic different iti different pln.ces. character of the earth and from a knowledge of 3. The density of the interior is greater than the composition of m~ny meteorites, that part, that of the known crust, for the n1ean density at least, of the earth's interior is composed of the earth ·as a whole is about 5.52, whereas largely of iron or of a nickel-iron alloy sin1ilar th~t of the crust is. about 2.79. 2 Th~ density to that which constitutes the iron meteorites. at or near the center 1nay possibly be set clown An objection cmnmonly made to this view is n.s of the order of 1nagnitude of 8 to 10, accord- that iron loses its magnetism at 785°, but this ing to the assmnptions that are 1nacle; or it objection is invalidated by our complete ig- 1nay conceivably be even higher. The prob- norance as to the n1agnetic behavior of- iron lmn is at present insoluble, because we know at the enorrnous pressures that exist· beneath nothing of the rate· of ~hH.nge in density with the outer crust. Furthermore, by analogy depth; nor do we know whether the density with 1neteorites, such a n1etallic core would be Vftries continuously with depth, or whether it cmnposed of a nickel-iron alloy, not of pure becmnes constant below a certain depth. iron; and it is well known that the nickel-iron l!. The earth as a whole acts in 1nany respects alloys possess n1agnetic properties that are as a 1nagnet, and as the rocks of the crust in very different from .those of iron or steel. The same is true of the n1anganese-iron alloys. It ' 'l'his introduction is taken bodily, except for some minor changes anti additions, from t\ paper by H. S.. Washington on "The chemistry . is thus quite Within the ~ounds Of reasonable or the earth's ernst," published in tho Journal of the Franklin Institute for December, 1920 (vol. 190, pp. 757-768). ·a Adams, L.I:I.,and Williamson,K D.,'rhecompressibilityofminerals ~In tho paper from which this introduction is taken tho density of and rocks at high pressures: Franklin Inst. Jour., vol. 195, pp. 475-529, tho crust was given as 2.77. Later computations give 2.79. See Wash- 1923. ington, H. S., Isostasy and rock density: Geol. Soc. America Bull., vol. 4 Williamson, E. D., and Adams, J_,, H., DollSity distribution iu tho 33, pp. 375-410, 1922. . earth: Washington Acad. Sci. Jour., vol. 13, p. 413, 1923. · 1
7. 2· THE COMPOSITION OF THE EARTH'S CRUST. possibility that the nickel-iron alloys which are that e-ventually solidifies as a rock is called here assuined to com pose a zone in th~ interior technically. the ''magma," a teru1 that. is in of the earth 1nay retain their magnetic- sus- frequent use in petrology. ceptibility even at the high temperatures that The magn1a cou1es up frou1 below; frorn what are assun1ed to exist at great depths; and this depth we do not know, though there is sou1e possibility is enhanced when we consider the reason for thinking that the places of its origin ·unknown influence of great pressure. are not very deep. ·Nor do we know whether it Leavipg the ·interior of the earth for the arises fro:tn the melting -of parts of the earth present, we ·1nay concentrate our attention on that are actually solid but potentially liquid the outer shell-the so-called ''crust "-which on relief of pressure, or whether it is in general is the only pa'rt of the earth that is direetiy derived from ((reservoirs" of liquid magrna. open to our study, and which has been com- The igneous u1agma is usually compared to a pared, with so1ne reason:, to a covering of slag Co:tnplex solution of salts in water. This COlll- or scoria over the interim:·. In dealing with. parison, which was first suggested by Bunsen this crust we shall consider prin1arily its in 1861, has been very fruitful of results in the che1nical cmnposition. study of the origin, fonnation, and character The thickness of this crust is unknown, is of igneous rocks. probably not unifonn, and is preswnably The Jnaginrt contains gases in solution, Jnuch indeterminate. · For purposes of computation as spring wat,er contains air in solutioi1, or as we may assume· an approximate thickness of the waters of 1nany mineral springs. contain io rniles (16 kilmneters), this being about the carbon dioxide, ·which escapes on relief of slun of the highest elevation of the land and pressure. the greatest .depth of the ocean. Of these 1nag1natic gases the one that has by This solid crust is n1ade up alrnost wholly far the greatest effect and that is generally the of igneous rock-that is, rock that has·solidified most abundant is the vapor of water. This va- fro1n a hot, liquid {" n1olten ") condition, either por fonns the greater part of tlie clouds that as ((plutonic" rock, found at different depths are given off by volcanoes in eruption and, with beneath the surface, as dikes or sills filling other gases, produces the spongelike structure crevices, or as lava flows .at the surface. seen in pumice and the cavities of vesicular Assmning a thickness of 10 n1iles, the cmnposi- .lavas; an effect caused by expansion, due to tion of the rocks of the crust is esti1nated to be relief of pressure in the lava as it reaches the about as follows: surface. Some glassy lavas contain a rather huge percet1tage of water, the 1nagma having Proportion of rocks of di.fferent classes in the earth's crust. solidified so rapidly as not to permit its escape, Igneous rocks ....................... 95 and inclusions of water and liquid carbon· Shales.............................. 4. 0 dioxide are visible in the crystals of 1nany Sandstones............................. . 75 Limestones ....................·....... . 25 gr.anites and other rocks. The presence of ·water in volcanic magmas has been doubted 100.00 by Brun and others 'vho followed him, but its The rnetaJnorphic rocks, such as gneiss and existence in lavas, especially in those of Kilauea, schist, are hereincluded with the igneous rocks. has been shown conclusively by the researches In a study of ·the chemistry of the crust as a of Day and Shepherd 5 and is made eviden~ by whole such relatively small masses as beds of practically every rock analysis and in other coal or deposits of salt and ore are negligible, ways, so that the existence of water in magn1as though their presence is significant, and the 1nay be regarded as one of the established coating of soil is also negligible. truths of the che1nistry of igneous rocks. Volcanic exhalations contain not only the GENERAL FEATURES OF IGNEOUS ROCKS. vapor of water but other gases-carbon dioxide, Igneous rocks, as has been said, are those that carbon monoxide, hydrogen chloride, sulphur have solidified from a state of fusion, or prefer- trioxide and dioxide,.hydrogen sulphide, carbon ably liquidity, for the term fusion implies a 6 Day, A. L., and ShepherQ., E. S., Water and volcanic activity: previously solid condition. The liquid m&tter. Geol. Soc. America Bull., vol. 24, P· 573, 1913.
8. THE COMPOSITION OF THE EARTH'S CRUST. 3 oxysulphide, hydrogen fluoride, ammonia,· a 1netasilicate of potasshun and ahuninum; 1nethanc and possibly other hydrocarbons, (9) 1nagnetite, ferroso:ferric oxide, and ilmenite, sulphur Vitpor, hydrogen, nitrogen, oxygen. oxide of iron n.nd titanimn; and (10) apatite, T.he presence of these gases in the magn1a a phosphate of · calciu1n containing a little lowers its solidification point, so that a ~ava, fluorine or chlorine. Magnetite and apatite are o.r1 coJning to the surface, 1nay be and usunJly is found in nearly all rocks, but seldo.m in .more I iquid at It te1nperature considerftbly below the thftn sn1nll a1nounts. fusing poii1t of the solid Tock Jonned .fron1 it, Certain kinds of igneous rocks contain other for during :its solidificn,tion n1uch of the clis- Jninern1s, such as the silicates sodn1ite, hai.iyne, solvecl gas is lost. Either because of this loss 1nelilite, zircon, and garnet, and the oxides or because of an increase in fluidity, or in SOlne tridynlite (a second :fonn of silica)' hmnatite, other wn.y that is .i'wt yet well understood, the chrOJnite, spinel, co!·unchun, and rutile. But gnses in the mn.gn1n. seeJn ~o promote the in a study of the igneous rocks that forni. the crystalli7.1tt:ion of minern.ls, so that they are whole crust of the earth these 1ninerals are often referred to n.s "Jnineralizers." S01ne of practically negligible. · Igneous rocks, then, these gases also play an iJnportn.nt p!1rt in the in general, looked at in the broadest way, are fo:nnation of n1n.ny ore bodies. fonned ahnost wholly of a very few silicates of Tho nutg.nHt on solidification generally :fonns alu:minu1n, iron, calcimn, 1nagnesiu.m, soclimn, n. mixture of minerals, suhstances of definite potassiu1n, and hydroxyl, with or without chemical co.mposition and physicn1 clutnt<;:ter, quartz (that is, excess of silicn.),.with s.mall just tts n. solution of salts in water (such as sea amounts of a phosphate and of free iron oxide, Witter) forms a mixture of crystals of snits ltnd and with or without traces of other constituents. ice on freo7.ing. Frequently the rnagnut. eools Son1e. of the essential 1ninerals einunerated too rapidly to permit complete crystallization abov~ (pyroxene, mnphibole, 1nica, olivine, and or (as with obsidian) too rapidly for any 1nagnetite) contain sJnall a1nounts of Jnanga- crystallization, so that the rock is cornposed nese and titanium. Fron1 such a general survey pn.rtly or wholly of glass. Such glassy rocks of the rock-forn1ing 1ninerals, then,· we obtain n.re found only as surface flows. a broad conception of the che.mical c01nposition of the earth's crust as a whole. MINERAL CONSTITUENTS OF IGNEOUS ROCKS. Another in1portant fact concerning the 1nin- It is a very sigi1ificant and striking fact that ern1s that fonn igneous rocks is that, with two alt:!wugh about a thousand different 1ninerals exceptions, any one of the1n 1nay occur in rocks nrc known, yet the nu1nber that co1npose over with any one or 1nore of the others. The only 99 per cent by weight of igneous rocks is very exceptions are that neither nephelite nor leucite s.mitll. Indeed, the n1inernls that are really is known to occur with quartz, and a partial essential to fonn igneous rocks number less exception is that olivine seldonl occurs with tlutn a dozen. · quartz and never in large amount. These essential' 1ninerals are (1) quartz, Of the ;minerals na1ned nearly all may be pres- silicon dioxide; (2) the feldspars, silicates of ent in a rock in 'Yidely varying proportions, and alu1nina and po~ash, soda, or lime, including each may be con1pletely absent. V'\T e know igne- the potassic orthoclase, the sodic albite, and ous rocks that are COlnposed entirely of quartz the calcic anorthite, with ison1orphous nlix- (northfieldite), feldspar (anortho~ite), pyroxene tures of these; (3) the pyroxenes, 1netasili- (websterite), a1nphibole (hornblendite), or oli- C11tes of Ci1lcimn, 1nagnesiu1n, and iron, sonle- vine (dunite), and ahnost entirely of nephelite ti.mes with ahuninu1n or sodimn; (4) the am- (congressite), leucite (italite), or nutgnetite phiboles, in chen1ical coniposition 1nuch like (some iron ores), and some rn.re igneous rocks the pyroxenes, but differing in crystal forn1 and are composed wholly of biotite. Of the essen- otherwise; (5) the 1nicas, alumino-silicates, tial rock n1inerals, only apatite does not fonn n1ostly the potassic 1nuscovite or the potas- the whole or aln1ost the whole of any known siun1-iron-1nagnesiun1 biotite, all containing igneous rock. . hydroxyl; (6). the olivines, orthosilicates of We also know some igneous rocks that are . iron and 111agnesium; (7) nephelite, an ortho- composed of two 1ninerals, 1nore that are COln- silicate of sodium and al~1ninu1n'; (8) leucite, posed of three, and many that are composed
9. 4 THE OOMPOSITION OF THE EAHTH'S CRUST. of 1110l'C [than:' three, with!th( widest possible silicate rocks! silica shows easily the highest variations in the proportions of almost all, ex- maximum and the .widest range, both in cept, as noted above, that quartz does not extremes and in the usual run of ·occurrence. occur with nephelite and leucite and that its A few igneous rocks are composed almost occurrence with olivine is rare .. entirely of quartz, and the highest percentages of silica· recorded for truly igneous rocks are CHEMICAL CONSTITUENTS OF IGNEOUS· ROCKS .. in one from Cl1mberland (England), where the Although, as we have seen·, most ignequs border facies of It granitic mass shows 96.16; rocks are composed of but few essential min- one from Massachusetts with 93.38; and one erals and consequently of but. few so-called from Arizona . with ·92.59. In general, how~ "1najot" oxides (in terms of which the chemi- ever, the percentage of silica ranges from about cal con1position of rocks is usually fonnulated) 75 to -about 34, and it drops to zero only in yet when we come to study the igneous rocks, some "magmatic" iron ores. In nearly all in detail we find that some rocks may contain rocks it is the. most abundant constituent. a considerable nu;rnber of che1nical constituents.- Veins Qf quartz are not. considered here, as. Altogethe1', about twenty-three are determined many of them are of nonigneous· origi:ri, in the and recorded i11 _rock analyses of the better ordinary sense of the word igneous . . class. Indeed, as has been saicl'hy Dr. W. F.. Alumina, which is almost 1nvariably the Hillebrand, the foremost analyst of rocks, "a next mos~ abundant constituent, reaches a sufficiently careful examination of these: maximum of about 60 per cent in some [igneous] rocks would show them to contain all corundum-bearing syenites from Canada and or nearly all, the known ele)nents; not neces, the Urals and has a general·range f1:01n about sarily all in a given rock, but n1ore than any- 20 to about. 10. It is wholly absent only in one has yet found." Proper study, therefore, the crmagmatic" ores and in some rocks that of the che;mistry of igneous rocks and their are composed largely or entirely of olivine . . che1nical· analysis, if it is to detern1ine all the The two oxides of. iron, of course, reach their constituents probably present, is evidently ~axima in such rocks as the iron ores already solnewhat difficult and complicated. · spoken· of, the highest figures recorded for About 13,000 chemical analyses of rocks hav.e Fe 20 3 being 88.41 (Sweden) and 62.39(0ntario), been ·made since their analysis. was first at- and the highest for FeO being 34.58 (Sweden) tmnpted, near the beginning of the nineteenth and 32.92 (Minnesota). Their general' range. century, and these analyses show us what are is from nearly 15 for each (FeO generally the chemical constituents of rocks and give us higher than Fe 20 3), and but little more than a good idea of their relative abundance and the that for both together in any one rock, to less ranges in their percentages.- than one-half of 1 per cent. . Iron is' found in. The most abundant or "major" constituents, all but a v~ry few rocks.· stated as oxides, in the order in which they are· Magnesia reaches its maximum in the sta.ted in the analyses, which is not quite the -almost purely olivine rocks (dunite) of North order of their abundance, are silica (Si0 2 ), Carolina, 48.58, and of New Zealand, 47.38, niu1nina (Al 20 3) , ferric oxide (Fe2 0 3) , ferrous but it ranges in general from about 25 to ·oxide (FeO), 1nagnesia (MgO), lime (CaO), much less . than 1 per cent. Lime is highest soda (N a 20) ,·_potash (K 20), and water (H 2 0). (22.52) in smne pyroxenites of the Urals, an_d These nine oxides together n1ake up about 98' almost as high (about 20) in the anorthosite of per cent of the igneous rocks, and all occur in Canada and elsewhere, but it ranges in general greater or less an1ount in practically every from about 1.5 per cent to nearly zero. rock, so- that the quantity of· each must be Of the two· alkalies, soda reach,es a maximum determined· in every rock analysis that makes of 19.48 in a. rare almost purely nephelite · the slightest pretense to good quality. rock from Canada and of 18.67 in another As the most abundant and essential rock from_ Turkestan, but its general range is from . minerals are either silica or silicates, and as about 15 per cent down to. nearly zero. It is all igneous rocks except. some -rare and small entirely" absent in only a few rocks. Potash bodies of iron ore of magmatic origin (smne of shows a somewhat ·smaller range· than soda, which may. not be igneous) are consequently· its maximum being _17.94 in an aln1ost purely
10. THE COMPOSITION OF THE EARTH's CRUST. 5 loucitic.lava(italite) from Italy, the next highest trace of it, although' this may be because of figure being 11.91 from ·Wyoming; but in the more delicate tests required for the other genert~.l it seldom gets. above 10 per cent, two. rn.nging from that down to zero. Its amount Mariganese, ·as manganous oxide (MnO), is is generally less than that of soda. ·present in practically every rock that has been As regards water, the last of the major analyzed, but its maximum is much lower than constituents, a few volcanic glasses, which are the maxima of titanium and phosphorus oxides. perfectly fresh and undecomposed, contain as Some of the high figures reported for· it are much as 10 per cent, and some fresh crystalline almost certainly due to analytical errors, and rocks contain fron1 3 to 5 per cent. Generally, the highest recorded trustworthy figures are however, a ro·ck that contains more than about 1.90 and 1.46, shown by two rocks from Bahia, 2 per cent of II20 has acquired the excess by Brazil. Its general range is from 0.3 per cent n.ltern,tion, and a few rocks contain no water. ·to almost zero. Effusive rocks contA.in a considerably.larger The· ·other 1ninor constituents that are ~tmount of water than the chemically equiva- readily determinable, Inany of which, indeed~ lent intrusive rocks. 11 are determined in good analyses, are car- The "minor" constituents are generally bon dioxide (00 2 ), zirconia (Zr0 2), chromiun1 found. in very smnJI. a-mounts, rarely more than sesquioxide (Cr20 3), vanadium sesquioxid.e 2 per cent for any one or as much as 5 per cent (V 20a), the "rare earths" ((Ce,Y) 2 0a), nickel for nll of them, in any one rock. Of these oxide (NiO), stront.ia (SrO), baria .(BaO), minor constituents, three are of special im- lithia (Li 20), sulphur as both sulphide (S) portn.ncc, partly because of their · ahnost and sulphur. trioxide ·(80 3), chlorine (01), eonstn,nt presence iwd partly because they are and fluorine (F). To these· 1night be added gencrnJly present in largest anlolmt. These boron, cobalt, copper, glucinum, ·lead, molyb- threc constituents n.re titanimn dioxide, phos- denmn, nitrogen, and zinc, the quantities , phorus pcntoxide, and mangnnous oxide, and of which, however, are gener~lly so very small in a ·good rock analysis all three should be . or the analytical diffi.cul~ies for the sepn.ration determined. of the t,rifling quantities. of the1n that occur are Titanium dioxide (1'i0 2) reaches a maximum . so great that their determination is rarely in some v:ery rare rocks from Virginia (69.67 attempted. u.nd 65.90) and Quebec (53.35), but 1ts per- The maxima and ranges of son1e of these centage is seldom· over 5 and generally ranges constituents may be briefly stated. Carbon fro:m about 2 to nearly zero. Probably not dioxide may be a component of a few n1inerals one· of the many rocks ~f all kinds that have (such as primary calcite and cancrinite) in . been analyzed contained no titanium; in some some unaltered rocks, but its presence is gen- rocks the quantity is very small, but in almost erally due to alteration by weathering or to the nll it is easily determi;nable. This fact is shown assimilation of limestone by the magma, as by the work of Hillebrand 7 and by the work, in the nephelite syenite of the I-Ialiburton area, as well, probably of every other experienced Ontario, 8 and probably the· alnoite of Alno. 0 analyst of rocks. There are, howeve~·, .some rocks in which the ·The n1a.ximum· for phosphorus· pentoxide calcite is apparently of primary origin; as a (P 20r,) is a little above 16 per cent in some trachyte of Mount Axpe, ·nea.r Bilbn.o (unpub.:. highly unusual rocks from Sweden and Vir- li~hed) and a calcite alaskite of Alaska. 10 The ginia thn.t are composed largely of apatite, Bilbao trachyte contains 7.69 per cent of 00 21 with which is found titaniferous magnetite or and in rocks with primary cap.crinite carbon rutile. In few rocks, however, is it above 3 dioxide may ~~each 1~70 per cent, but in gene:ral p"er cent, and its general range is fron1 about s Adams, F. D.,1 and Barlow, A. E., Geology of the Haliburton and 0.5 per cent to. zero. It does not seem to be Bancroft areas, Province of Ontario: Canada Gcol. Survey :Mcm. 6, ·JL present so constantly as titanium or manganese, 233, 1910. 9 Tomebohm, A. E., Melilithbasalt fran Alno: Geol. Foren. Stockholm ns one occasionally meets a tock that sho'ws no Forhandl., vol. 6, p. 240, 1882. Hogbom, A. E., Ueber das Nephclin- e Wnshington, H. S., in :Allen, E. T., Chemical aspects of volcanism: syenitgebiet auf der Insel Alno: Idem, vol. 17, p. 234, 1895. :L<'mukliu lust. Jour., vol. 19:3, p. 41, 1922. 10 Wright, C. W., Geology and ore deposits of Copper Mountain and 7 Hillebrand, W. F., The analysis of silicate and carbonate rocks: U.S. Kasaan Peninsula, Alaska: U. S. Geol. Survey Prof. Paper frl, p. 81, Gcol. Survey Bull. 700, p. 25, 1919. 1915.
11. 6 THB COMPOSITION OF THE EAHTH 's Cl{,UST .. carbon dioxide is to be regarded as a 111easure possibly of magmatic origin, and it is reported of the alteration of the rock by weathering or as forming between 2 and 3 per cent of som~ otherwise. Bragger 11 has recently des-cribed a undoubtedly igneous rocks from Baden .. But series of intrusive rocks from Norway that these are highly exceptional, for about 0.5 contain large ~mounts (up to 80 per cent) of may· be taken as its usual maximum. It is calcite, which he regards as primary. generally quite absent. Vanadium sesqui- Zirconia is much less abundant than the . oxide is widely diffused in the igneous rocks closely related titnnic oxide, and, although it but principally in the :r:nore fernie kinds. 'l'he reaches a Inaximum of nearly 5 per cent in oxides of the rare earth metals, chiefly ceria son1e rocks in Greenland, its amou1~t is rarely and yttria, reach a maximum of 1. 79 in a rare n~ore than 1 per cent and is usually much type of rock from Madras, 0.6 in one from less, and it is quite absent from most rocks. It Sweden, and 0.4 in one from the islet of Rockall, forms one of the most striking illustrations of but the usual maximum is only one 'or two the concomitant occurrence o:f elements in tenths of 1 per cent. . They are less of ten different kinds of rocks; · determined than they shoulCl be. Nickel oxide . Baria and strontia are ver.y commo.nly is present in some rocks up. to about 0.2 per found, although they are ~eldom determined cent. The maximum ·amount of each of the in analyses 1nade outside of the United States, other minor constituents may be placed at Canada, Great Britain, and Australia. ·The not over 0.2 per cent,· and they are almost amount of haria is almost invariably much always found only as one or two tenths or still greater than that of strontia, this being an more often: as hundredths of 1 per cent, or exception to a general rule as to the occurrence they are absent. Indeed, :for most of the of related elements. They both reach their minor con~tituents the quantities usually maxima in certain exceptional highly potassic yielded by analysis are so small as to be· rocks of Wyoming, which contain about 1 per significant _only as to their actual presence or cent of haria and 0.3 of strontia; though usually absence. the percentage of baria amounts to only a few. AVERAGE COMPOSITION OF IGNEOUS ROCKS. tenths of 1 per cent and that of strontia is Ineasurable in hundredths of 1 per cent. AVERAGES COMPUTE:p. Sulphur is present as sulphide in amounts More than 30 years ago the senior author of up to about 9 per cent in a peculiar pyrrhotite- this paper 12 atte1npted to estin1ate the relative bearing rock from Maine and probably in abundance of· the chemical eleme~ts by aver- similar amounts in some sulphide ore bodies of aging analyses of eer.tain classes. This esti- Inagmatic origin in Norway, which have not mate included the lithosphere (as represented been fully investigated, but its amount is by igneous rocks), the ocean, and the atmos- seldom over 1 or 2 per cent, and it is usually pher'e. For ·the igneous rocks 880 analyses :found in tenths of 1 per cent. 'l'he highest were tak~n, of which 207 were made in the . figures for sulphur trioxide are about 2.5 per laboratories of the United States Geological cent, in rocks from·Apulia and Cameroon, and Survey and 673 were made· at other places. percentages somewhat lower in rocks from These were divided into groups, each group Tahiti, but sulphur :trioxide is usually present representing a definite geographic area, and onlyintenthsorhundredthsoflpercent. Much it was shown that these groups of averages the same can be said of chlorine, the high.est agreed fairly well with one another. That is, :figures foi· which are those for a rock from 1 the con1position of the earth's crust as thus Turkestan (ab.out 7), one from Quebec (4.4 7), determined is approxin1ately the san1e in its and one from French Guineac (2.80); It is different parts, provided large enough areas present in many rocks, especially lavas, but are taken to eliminate purely local variatio~s. only in a few tenths or hundredths of 1 per. Since this first atten1pt was 1nade other aver- cent. ages have been computed with more abundant Chromium sesquioxide is present up to about data and by ·different methods. All these 4 per cent in some o~es from Greece, which are averages may be briefl} summarized as follows: 7 u Brogger, W. C., Die Eruptivgesteinc des Kristianiagcbietcs: IV, 12 Clarke, lf. W., The relative abrmdance of the chemical elemen'ts: Das Fengebiet in 'l'elcmark, Norwcgen, Kristiania; 1921. Philos. ·soc. Washington Bull., vol. ll, p. 135, 1889.
12. THJ 1. The original n.verage, 1nade as described few analyses available. This assun1ption n1ay abov-e. or n1ay not be true; for son1e countries it 2. An average of all the analyses of igneous would appear to be decidedly erroneous. rocks, partial or con1plete, ~nade in the lab- Still another objection is that the data con- oratories of the United States Geological sidered include an undue proportion of analyses Survey up to October I, 1918/3 Son1e other of the more·" interesting" types of rock and n.verages of the srune order were made but that, on the other hand, the 1nore ordinary need not be'n1entioned here. ·' · (and hence less "interesting.") kinds are not 3. An average, con1puted by Harker/ 4 of sufficiently represented. Although this ohjec- 536 analyses of igneous rocks from British tion is valid, yet it would semn not to be so localities. Many of these analyses were in- serious as has been asserted, fo.r the analyses COinplete, especially with regard to titanimn, of rocks of the satellitic dikes and other small phosphorus, ru1d 1nanganese. bodies that are n1ost likely to furnish "interest- 4. An n.verage, c01nputed by Washington,t 5 ing" types tend to c01nplmnent one another of 1,811 n.nalyses fr01n all parts of the world. and so to give approxiinately the true mean. To this method of averaging serious objec- Again, -it is by no means uncommon that tions have been raised. All analyses are given analyses of the Jnore abundant kinds of rocks equal weight, without regard to the areas occu- in a given district have been n1ade and none pied by the rocks analy7.ed, and therefore to of the sn1all dikes. Furthermore, the more their relativ-e abundance. One rock, say- a subsilicic rocks, mru1y of which ar~ of the granite, is exceedingl)T abundant; another may more "interesting" types, are most subject to be represented by· one small dike. The in- alteration, so that the analyses of many of equality is obvious, but what. does it really them would be excluded fron1 the data here signify~ In the first place, the relatively in- used, which include only analyses of fresh, · significant rocks vary in composition from unaltered rocks. persilicic to subsilicic just as the most abundant Averages based on a quite different n1ethod rocks do; In the average these variations tend of computati.on have also been made. Daly 16 to offset one another and so to give an ap- has made separate averages of many rock proxinlately true 1nean. Furthermore, the types, and he shows that the mean composition surface exposure of a rock is no certain measure of an average granite combined with that of of its real volume and mass, for a small ex- an aver.age ·basalt is almost identical with the posure n1ay be merely the peak or crest of a average given by Washington. Mead,t7 apply- large subterranean body and a large exposure ing a peculiar graphic method to some of may represent only a thin layer. Daly's · data, concludes that a mixture of 65 Another obj.ection is that the igneous rocks per cent of the average granite· with 35 per of large parts of the earth are quite or almost cent of the average basalt will have a conlpo- . unknown, at least in an exact chemical way. sition very close to the general average of all This is true of the interiors of several of the igneous rocks as computed by Clarke. A continents, notably Asia, Africa, South Amer- similar result is reached by Loewinson-Lessing/ 8 ica, and Australia. The rocks of many .coun- who regards the crust of the earth as derived tries, also, are inadequately represented by· from· two fundamental n1agmas, one granitic chen1ical analyses. Examples are China, india, and one gabbroic. These are supposed to Asia Minor, Arabia, Egypt, Spain, Brazil, have existed in about equal proportions, and Argentina, Chile, Central America, Mexico, their mean composition is nearly that here ru1d northern Canada. In making the average found for all igneous rocks. analysis of the rocks of all such countries it is Such averages as these, however, are arbi- tacitly assumed that the. whole of the area trary, ru1d the correspondences would seem to (continent or country) is represe~ted by the be in reality fortuitous. The same result as, ta Clarke, F. W., 1'ho data of geochemistry,4th ed.: U.S. Geol. Survey te Daly, R. A., Average chemical composition of igneous-rock types: Bull. 695, p. 26, 1920. Am. Acad. Arts and Sci.. Proc., vol. 45, p. 211, 1910; and Igneous rocks 14 Harker, Alfred, •rertiary igneous rocks of the Isle of Skye: United and their origin, pp. 19-46, 168-170, New York, 1914. Kl..ngdpm Geol. Survey Mem., p. 416, 1904. 17 Mead, W. J., The average igneous rock: Jour. Geology, vol. 22, p. · ~~Washington, H. S., Chemical analyses of igneous rocks, published 772,1914. from 1884 to 1900, with a critical discussion of the character and use or IB Lciewinson-Lessing, F., The fundamental problems or petrogenesis: analyses: U.S. Geol. Survey Prof. Paper 14, p. lOG, 1903. Geol. Mag., new ser., decade 5, vol. 8, p. 248, 1911.
13. '.8 THE cOMPOSITION OF THE EARTH's CRUST. let us say, that of Daly may .be arrived at by igneous areas in the Appalachian. and Cor- an infinite. variety of combinations of· various dilleran regions, as shown in the folios of the kinds of magma. Thus, if one starts with a United States Geological e,urvey, and then mixture of 10 parts of average granite and 1 combin.ed the results so determined to form a · part of average basalt, the high silicity and general mean. Theoretically, Knopf's method other peculiar chemical features of the granite . is· highly plausible, and it would be the best (as contrasted with those of the basalt) may be if it> could be generally utilized, .but for the counterbalanced by the introduction of appro- purpose of calculating the average of a shell priate portions of, say, perido~ite,. pyroxenite, of any given depth it is· open to two. serious and anorthosite; or these, with possibly a. objections; it assigns as much weight' to . little nephelit.e syenite, might replace the comparatively thin lava flows as to exposures basalt. entirely and give the same result; and of batholithic masses, a~d also it takes no so ·on, ad infinitum. Indeed, the averages cognizance of the igneous rocks that underlie that have been obtained by the method here the. sedimentary beds. But in spite of these adopted illustrate the results of such a proce- objections this method is a distinct· improve- dure. .By judicious qualitative and quantita- ment on the original method of obtaining an tive selection of data one may ·arrive at any earth or regional average. . "average rock'' desired. Such averages, based ~ow, if we inClude the estimates made by on arbitrarily adjusted selection of data, mean Daly, Loewinson-Lessing, and Knopf, there 'little, and they would seem to be an illogical are seven averages to be compared. Mead's and unsafe basis for generalizations. as to the figures are omitted, for the reason that he average composition of rocks. It is still more neglected titaniun1 and phosphorus, which illogical to argue from the coincidence that a appear in all the others. Manganese, which given mixture has the same composition as is given in the original computations, i~ re- that of the general magma, and that th~ earth jected here but will be ~onsidered later, along magma js made up of such a mixture of with the oth~r . minor constituents of the submagmas. ·igneous rocks. Water is also left out of ac- In order to meet the obJ. ection · mentioned count for the present, because in too many analyses there w~s no discrimination between 19 above Knopf took Daly's averages ·for the various rock types ·and weighted each. type extraneous and essential water. In I'able 1 by its area, as determined by Daly for the the averages are all recalculated to 100 per ts Knopf, Adolph, The composition of the average igneous rock: Jour. Geology, vol. 24, p. 620, 1916. · cent. TABLE I.-Previously published estimates of the average composition of igneous rocks. - - - - - 1 - - l - - - : - - : - - l --!-..,......-1--·1, ___1~_ 3 6 7 .. Si0 2 ••••••••••••••••••••••••••••• ·••••• 59. 97 61. 69 60. 76 58. 96 . 60.18 59. 731 G2. 52 . i:§~: ~- ~ -~ ~ ~: ::::::::::::::::·: ::::::::. 15.39 15. 47 15.87 15.99 15. 55 16. 53 . 15. 93 '4.03 2. 71 4.92 3.37 3. 56 2. 42 2. 95 3.56 3.54 2. 78 3. 93 4. 07 3. 86 3. 3Q 3.87 3.82 3.89 3. 57 4,40 3.01 ~:8. ~ .· :. ~ :::::::::::::::::::::::::::: 4. 60 5.41 4.98 4. 97 5. 28 5. 63. 6. 67 5.·14 3. 29 2.95' 3.45 ~:0~ . ·~ ....·~ ~ ·.: . ~ ::::::::::~ :::::::::· : 3.28 3.48 3.28 3.96 2.97 3. 14 2. 85 3. 20 2. 89 2. 50 2. 69 Ti0 2 ••••••••••••••• ; •••••••••••••••• . 56 . 82 . 53 1. 05 . 90 .. 68. . 74 P205 .... ·.. ·.· ....................... . . 23 . 30 . 22 . 37 . 36 . 26 . 27 100.00 100.00 100.00 100. 00 . 100. 00 100.00 100.00 1. Clarke, 1889. 2. Clarke, United States rocks, 1918. 3. Harker, British rocks; 1904. 4. Washington, 1903. 5. Daly, granite-basalt average, 1914. 6.. Loewinson-Lessing, granite-gabbro average, 1911. 7. Knopf, Appalachian and Cordilleran rocks, 1916.
14. THE COMPOS·ITION OF THE EARTH'S CRUST. 9 Colmnns 2 and 4 include the largest nu1nber better results ought to be obtainable, and as of trustworthy analyses. They are, however, the data are now at hand. theil· discussion by not strictly comparable. Column 4 includes geographic groups will be atten1pted. analyses of rocks from all over the earth and In a work compiled by the junior author 20 :includes only analyses that were nominally more than ~,600 analyses of igneous rocks are· comi)lete and that \y-ere made in. many labora- brought· togethe:r and classified, and of these tories by somewlul.t diverse methods. Col- n1ore than 5,000 are available for present use. umn 2 includes analyses of rocks from the In the· work just cited the analyses are rated United States alone, and the analyses are the as superior, incmnplete, and inferior. 21 The honlogeneous work of one laborn.tory; more- inferior analyses· we 1nay disregard; their inferi- over, it includes n1any partial analyses. In ority is due to lack of co1npleteness, lack of analyses of the simpler salic rocks detennina- ·accuracy, or to both together. None of then1 tions of silica,' lilne, and alkalies are sometimes have been included in the data here considered. n.ll that are needed for petrographic purposes. The superior analyses are those which are The fernie rocks are chemically and nlineral- satisfactory (according to certain standards ogically more cmnplex and require more com- adopted) as to both con1pleteness and accuracy plete analyses. The partial analyses, there- and which are therefo~·e dee1ned to be usable fore, represent chiefly salic rocks, and their. if the rock analyzed is unaltered. None of the inclusion in the average tends to raise the pro- superior analyses of altered rocks or of tuft's portion of silica and to lower the proportion .have been used. There is a s1nall group of of the other oxides. The salic rocks, however, analyses which ru:e incmnplete as to the deter- n.re tbe most abundant, and so the higher figure ruination of some one constituent but which ru·e for silica given in column 2 ·seems to be the otherwise superior. These can be used, for n1ore probable. they are incomplete only as regru·ds certain Though 2 and 4 are alike in their general details .that do not affect the percentages of features, yet there are some differences notable, the other constituents. The analyses con- especiallY. in the silica. In this respect 4 sidered nun1ber 5,159, only 302 of which are re.sen1bles 1, which also includes many rocks slightly incomplete. The effort has been from outside the United States. Stich differ- made to base averages on good data, the num- ences are due to regional peculiarities. Harker's ber of the data used being considered secondary British average~ as computed here free from to their quality. water and manganese, differs consider.ably In making the averages, .determinations of from 'Pis complete computation, which is given hygroscopic or unessential water (H 20-) have on page 54. The two averages of Daly and been rejected. If a constituent is definitely Loewinson-Lessing hav·e already been discussed reported as absent, it is given zero value in the very briefly; they would seem to be more of summations. A "trace" of any constituent interest as coincidences than as of. scientific has been counted· as 0.01 per cent. The tenn value. The very high value for silica and "trace," it ·should be noted, .is much misused. some other minor fe.atures shown by I{nopf's It frequently means only that the analyst has· average (column 7) arise from his selection of reason to think that the constituent is present analyses from only two mountainous areas in but has neglected. to ·deter1nine its an:iount. tJ;le United States, the average~. of whose r.ocks As applied to all constituents except lithium, a show these special features in marked degree. the presence of which is usually ascertained In spite of all divergences, however, the spectroscopically, the amounts covered by seven averages are in fair agreement in their "trace" ru·e more than 0.01 per cent, which is· essential features. An absolute agreement is, the usual limit of weighing. In the great of course, not to be expected. · The different majority of . analyses "trace" means one or ~stimates give silnilar percentages to. the impor- more tenths of 1 per cent and no.t uncommonly, tant constituents of the rocks and so encourage especially as to titanium, one-half to possibly us to believe that the method of averaging here 2u Washington, H. S., Chemical analyse~ or lgileous rocks published adopted is truly representative of the rocks of rrom1884 to 1913, inclusive: u.S. Geol. Survey Prof. Paper 99, 1917. , 21 For a discussion or the "rating" or analyses, the standards or quality . the earth's crust. With larger masses of data, adopted, aud the division or the collection or analyses into "superior" however, and ,,;ith wider geographic range,. and "inferior" parts, see pp. 18-28 or Professional Paper 99.
15. 10 THE COMPOSITION OF THE EARTH's CRUST. 5 per cent. The procedure adopted therefore I therefore yields results 'that are· practically gives .figures that are too low; that of count- free frmn error, at least so far as the data at ing 0.1 per cent for "trace" would have been hand are concerned. better and would have allo.wed for the 1nany In considering the "minor" constituents, nondeterininations. however, from titanium dioxide down to the end of the list, the conditions are quite different. METHODS OF AVERAGING. In many analyses some o.r all of these .con- Before the averages.compute~ are given it is stituents were not determined, yet their omis- necessary to consider a very important point- sion does not ·exclude "the incomplete analyses that is, the method of averaging adopted. so rigorously from the "superior" class as In practically all the analyses that served as the would the omission of the "major" constit- fundamental data, all the niajor constituents uents; out determinations of titanium, phos- were determined. These include silica, alumina, phorus, and manganese oxides are req~isite for ferric oxide, ferrous oxide, magnesia, li1ne, the best ratings and should always be made in soda, potash, and water. Indeed, the deter- analyses of rocks in which they are abundant. mination of all these was essential to admission Furthermore, the number of determinations of of an analysis to the grade· of "superior," the minor constituents differ widely, as will be the grade to which attention was confined. seen in Table 2, which presents the data con- In only a very few of the analyses used were cerning the most important minor constituents there exceptions to this rule. They included shown by the analyses here used. Titanium, a few "incomplete, but otherwise superior·" phosphorus, and manganese were. determined analyses, in which the iron oxides had not in mo§t of the analyses, sulphur and carbon been separately determined, and also a few dioxide. in many, but· the other minor con-: analyses of ultrafemic rocks (such as dunite) stituents we:re neglected in the great majority in which the alkalies were not determined. of the analyses made in all countries, especially The n1ethod adopted for obtaining the averages those made in continental Europe. Baria, for for these major constituents-that of dividing example, was determined in few analyses except the sum of the amounts of any given constit- those made in the United States, Great Britain, uent by the number of its determinations- and Australia. . TABLE 2.~Average percentages of th~ 1ninor constituents shown by 5,159 analyses. 2 . 3' 6 . 3, 926 4, 724.03 1. 20 0.91 1. 05 ~~d:_:~::_:_:~: :::::::::::::::::::::::::::::::::::: 76:0 3,.375 65.4 1,225.35 ·. 36 . 24 . 30 3,016 60.0 481. 77 .16 . 09 .125 Sa ...........................•................. 1,621 31.4 130.38 . 08 . 025 . 052 258.94 .199 .05 .102 .~~l;. ~ :_.. ·.: :::::::.: ::::::::::::::::': ::::::::::::: . 1,319 25.6 1,160 20.0 104.29 .09 . 02 . 055 . 01. ........................................... ·.. . 857 16.6 82.55 . 096 . 002 . 048 SrO .......... .- ................................ . 802 15;6 30.23 . 038 . 006 . 022 Zr0 2 •••••••••••••••••••• ·• • • • • • • • • • • • • • • • • • • • • • • • • 631. 12.2 44.41 . 07 . 009 . 039 Cr2 0 3 •••••••.•.••••••• , ••••.•••••.• : . . • • • • . • . . • . 477 9.2 48. 73 .102 . 009 . 055 Li20 ..................... : .· ......... ~ ............ . 449 8. 7 6.13 . 014 . 001 . 007 NiO ....................................... ~ .... . 421 8.0 19.57 . 046 . 004 . 025 F .............................. ; ............... . 223 4.3 33.99 .15 . 006 . 078 200 3.9 12:37 . 061 . 002 . 026 ~~~~---_·::::::::::::::::::::·::::::::::::::::::::: . 169 3.3 2.98 . 018 .001 ·. 010 a Includes S03 .. 1. Number of actual determinations. • 2. Percentage of determinations in whole number of analyses considered. 3. Sum of the percentages. <:::> · 4. Sum divided by the number of determinations. (Second method.) 5. Sum divided ·by-the whole number of analyses considered; that is, 5,159. (First method.) 6. Mean of averages 4 and 5. (Third method.)· ·
16. THE COMPOSITION OF THE EARTH's CRUST. · 11 There are three methods of averaging these so ·that a mean between them will probably lninor constituents. In the oldest method, be nearer the truth than either of them. • which was that originally used by the senior Although this conclusion may be true with ·re- author,22 the sum .of the percentages of any spect to some constituents, yet it is unquestion- given constituent is divided by the whole ably erroneous with respect to others. The nun1ber of analyses, it being assumed that a method is fundamentally faulty in bringing in constituent is not present if it is not reported elements that are unknown and arbitr&ry. or, in other words, that each analysis is com- It would appear, then, that the second plete. This assumption, however, is not justi- method, that of dividing the slim of the per- :fi.ed, and the method will give too low results centages by the number of determinations, in for the minor constituents and too high re- which 'we use only data actually known and suits for the major ones. avoid the introduction of the unknown arid the In the second method, which was used later arbitrary, is the most logical and the safest and by both the senior and the junior author the one which ·should give the most reliable separately, 23 the sum of the percentages of averages. It may be said that, even in dea~ing each constituent is divided by the number of with those constituents for which the third analyses in which it was determined. This method may yield possibly better approxima- methocl deals only with the data actua~ly at tions to the truth, the better and safer course hand, the data for the n1inor constituents being .is to use the second method, with the admission treated just as those for the major constituents that some constituent may be a little high, are treated. The data used are the only per- rather than to introduce the admitted uncer- tinent data known and are therefore the only tainty of the third method. We may leave data that can properly be used for this purpose. the correction of the departure from the truth The only assumption }llade in this method is to the future accumulation of more data. that the average amount of any constituent The matter is somewhat. more complex, represents the average amount of this con- however, than· it may appear to be at first stituent in all the analyses in which it was not sight; When we consider the data for th~ determined. This assumption, however, is several minor constituents in the light of our also ma.de· iri any method of determining an knowledge of the chemical analyses of rocks average and is fundamental to the jdea of and the quantitative and selective distribu- making averages. It may or may not be true, tion of the various elements among rocks of and though it is quite valid as to some of the ·different ·general chemical character, we find constituents it is less valid as to other~, and it that, besides the number of determinations, fi.trnishes an average based solely on known another factor should be taken into account in data. ascertaining correctly the averages of the minor In the third lnethod the average amount of constituents .. This factor is the unequal or n.ny constituent is the mean between those selective distribution of these elements among determined by the two other methods. This magmas of different general cheinical character. lnethod is based on the justifiable assumption Considered frOln this point of view the minor {or rather the knowledge) that the first method constituents may be referred to several dif- yields results that are too low and the possibly ferent categories., · justifiable assumption that ·the second method Some of them are .present to a greater or yields or lnay yield results that are too high, .less amount and one easily determinable by 22Clarke,F. W.,Relativoabundauceofthechomicalolements: Philos. :analysis in a_ll Or almost all igneOUS rocks, Soc. Washington Bull., vol. 11, ·p. 131, 1889; Report of work done in the '\vhether high 'or fow in silica, whether notably division of chemistry and physics, mainly during tho fiscal year.1889-90: u.s. Goo!. Survey Bull. 78, p. 3~, 1891; Analyses or rocks, with a chapter high in alkalies, lime, magnesia, iron, and so on analytical methods: Bull. 148, p. 13, 1897; Analyses of rocks: Bull. on. This is preeminently true of titanium, 168, p. 15, 1900. Also Harker, Alfred, On tho average composition or British igneous rocks: Geol. '?lfag., new ser., decade 4, vol. 6, p. 220, 1899, phosphorus, and manganese, the minor con- and 'L'ertiary igneous rocks of Skye, p. 416, 1904.' · ' 2s Washington, H. S., Chemical analyses of igneous rocks: U.S. Geol. stituents which are almost. always present in Survey Prof. Paper 14, p. 106, 1903. Clarke, F. W., Analyses or rocks: greatest amount, and in a modified degree of U. S. Goo!. Survey Bull. 228, p. 16, 1904; Analyses of rocks and minerals: Bull. 419, p. 6, 1910; Analyses of rocks and minerals: Bull. 591, p.18, 1915. some others. Other minor constituents are 63359-24--2
+1-315-636-4485
+91 98151-74050 
