The geochemistry of Darwin Glass

The geochemistry of Darwin Glass S. R. TAYLOR* alld M. SOLOM~~;~ (Hec:riwl16July

1963)

Abstract-Pieces of Darwin Glass, a frothy siliceous glass occur in soil overlying sandstones, limestones and volcanic rocks of Cambrian to Silurian age in \I’est,ern Tasmania. The refractive index averages 1.480. No radiogenic argon was detected, indicabing a maximum possible age of lo6 years. The glass is not uniform in composition, and at least, two distinct, but closely related groups exist. The average chemical composition is as follows: Si, 40.0% (85.62% Cr, 115ppm; Fe*, 1.63% (2.34% Fe,O,); SiO,); Al, 3.52% (6.64% Al,O,); Gn, 16ppm; Ti, 3520 ppm (0.59% TiO,); Ni, 122 ppm; Co, <3 Mg, 0.58% (0.96% MgO); Li, 3.7 ppm; ppm (24 ppm in Group I); Cu, 10 ppm; V, 29 ppm; Fe2+, 1.29% (1.66% FeO); Total Fe, SC, 4.1 ppm; Y, 30ppm; Na, 370ppm (0.051% Na,O); 2.57%; Zr, 390 ppm; Mn, 135ppm; Ba, 340 ppm; Ca, 740 ppm (0.10% CaO); Sr, 14 ppm; Pb, 7.6 ppm; K, 1,500/ (1.810/e K,O); Rb, 78 ppm; Cs, 3.6 ppm. The analytical methods are given, with values for the precision, A detailed geochemical comparison is made and the standard samples used for calibration. with terrestrial rocks, impact glasses, australites, a.nd with the adjacent country rocks. Darwin The rhemical composition does not Glass is not formed by terrestrial igneous processes. resemble that of t,he underlying rocks, but bears a close resemblance (except for nickel) to that of an argillaceous sand&one. The relative abundances of the alkali elements indicate that. little selective loss of the cations considered, has occurred during melting, and the present, Darwin Glass is not related to the composition is close to that of t,he parent material. anstralites. It closely resembles terrestrial impact) glasses, 1)articularly in the concentsation of nickel a,ntl t,he Fe/Ni ratios.

INTRODUCTION IN THE:year 1910,M. Donoghue, field assistant to L. K. Ward, Government Geologist, first directed scientific attention to the occurrence of a frothy siliceous glass from the eastern slopes of Mt. Darwin, Western Tasmania. The glass was first described, with two analyses (Table l), by SUESS (1914), to whom samples were sent. HILLS (1915) gave a full account of the discovery and location of the glass which was named “Queenstownite”, from the nearby town: by Suess. Later workers have adopted the name Darwin Glass. $ DAVID et al. (1927) gave a further description of the locality and occurrence, with two new analyses (Table 1) by Ampt. These workers considered Darwin Glass to be an aberrant type of tektite. Later workers (SPENCER, 1933, 1939; CONDER, 1934; PREUSS, 1936; S~ESS, 1935; EHMANN, 1960) classified Darwin Glass as an impactite, formed by the fusion of siliceous sediments during meteorite impact. PREUSS (1935) found an Fe/N ratio of 57 and EHMANN (1960) gave a value for the Fe/Ni rat’io in Darwin Glass of 99. These values are typical of terrestrial impact glasses. * Uepartment of Geophysics, Institute of Advanced St,lldies, -4tlstralian National

University. t Department of Geology, University of Tasmania. $ Both terms are subject to confusion. Queenstown is a common name in areas of British settlement. Mt,. Darwin is named aft’er Charles Darwin, \vho n-as one of the first scientists to inspect an australite, during the visit of HMR Beagle to Anstralia in 1835. There is no record that hr saw, or knew of the existence of Dar\vin Glass. 471

s.

472

1t.

TAYLOR and 31. s01,0u0n-

BAKER and GASKIN (1946) considered that the origin of Darwin Glass was uncert8ain. A principal argument against a meteoritic impact origin is the failure to identify an impact crater in the vicinity. BAKER (1959)SuggeSts ailOrigin by fusion of siliceous material in burning peat horizons. There exists some confusion about t,he authenticity of specimens, and itispossible that some smelter slag and Tertiary tachylite have been accidentally distributed as Darwin Glass.

-

SiO, Al‘p, Fe&I3 Fe0 MgO cao Sap K,O TiO, ZrOz &,(I, Xi0 COCI J&o ST.{f M& C’O, I?,(), SO, H,O

1 88.76 6.13 -.. 1.24 0.58 0.17 0.13 1.36 1.24

2

3

4

89.81 6.21 0.26 0.90 0.73 _ .

86.34 7.82 0.63 2.08 0.92

0.01 1.05 0.86 -_ --

O-l:? 0.87 0.52 0.11 nil nil nil nil nil nil nil

87.00 8@3 0.19 1.93 0.82 nil 0.14 0.99 0.51

OG

--_

---

tr

tr

. .99.83

99.61

tr

tr nil nil nil nil nil nil nil nil

nil

nil

0.46 99.95

0.36 99.94

1. Olive green glass, Darwin. Anal. E. Ludwig (SUSSS, 1914). 2. Dirty white glass, Darwin. Anal. E. Ludwig (SUESS, 1914). 3. Smoke grey glass, Darwin. Anal. G. A. Ampt (DAVID et nE., 1927). 4. Pale green-grey

glass, Darnin.

Anal.

G. A. Ampt

(Davrn

et al., 1927). -: not, determined. nil: not detected. tr: Iracr.

The present study was undertaken to investigate the geology and geochemistry of Darwin Glass, with particular reference to trace element abundances, and the relation to Lhc adjacent country rock. A prelimina,ry account of this work has been given by TAYLOR and SOLOMON (1962). DISTRIRUTIQN

AND

GEOLOGICAL

LOCATIOE

The Glass is abundant’ on the east slopes of Ten Mile Hill which lies 3 miles sout,h of Darwin and 12 miles SSE of Queenstown (Fig. 1). It occurs sparsely in scattered artas wibhin a zone .$-Imile wide that extends north from Ten Mile Hill for about

The geochemistry of I>aru?n Glass

4i3

(i miles, the hill being on the west side of the zone. Glass probably occurs to the south also but dense vegetation prohibits further observations. This zone is immediately east of the West Coast Range and in a relatively lowlying area, in which altitudes vary from 600 to 1500 ft above sea level. The Glass fragments are scattered in the peaty soil cover which varies from a few inches to

Fig. 1. Sketch map of Tasmania showing Daruin Glass localky. 18 in. in thickness. Many pieces appear broken but there is little evidence of streamrounding. The shallow depth of occurrence and the lack of abrasion indicate that the Glass is no more than a few thousand years old and formed or fell on a surface that has since been only slightly modified. Within the zone of occurrence there are four rock types, steeply dipping and outcropping in north-south strips (pa.rallel to the zone of distribution). From west to east across the zone these are: 1. Xt. Reed volcanics (Cambrian). Sheared volcanics. somewhat chlorit,ized Small quartz veins are common. and sericitized, mainly intermediate in composition. 2. 01~2 sandstone (Ordovician). Pink or grey, very coarse quartz sandstones and fine quartz conglomerates. Composed of quartz (~95 per cent) grains, slightly recrystallized and strained, subangular in section and between 0.5 and 1.0mm tliameter (coarser than Crotty sandstone). Quartz grains generally have thin. irregular fihns of hematite (1) and a few separate iron oxide grains are also present. C)ther minerals, rare in occurrence, are muscovite, green tourmaline and zircon (in order of abundance). No chromite was identified but some grains are indicated hy the analyses. dluminium chromite, containing about 55 per cent Cr,O, has been found in the Owen Sandstone. 3. Gordon limPstone (Ordovicha). Dark grey fossiliferous limestone and dark grey shale.

4. Crottt~ su~dst~r~~ (S~l~r~(l~). Pale grey eoa.rse quartz sandstone. Composed of quartz (about 95 per cent) grains, slightly recrystallized and strained, subangular in section and between 0.3 and 0.1 mm in diameter. The sparse matrix is either quartz (partly as overgrowths 0x1 grains) or sericitic material. A few flakes of muscovite are found and rare rounded grains of green tourmaline, zircon and rutile (in order of abundance). Only a few grains of opaque minerals occur and these are (There is no sign of chromite t,hough chemical analyses indica’te probably limonitic. the existence of a grain or two per thin section.) Darwin Glass is found on each of these formations, but mainly on t,he Owen Conglomerate and Crotty Sandstone, which underlie most of the area. Glass has been found on the volcanics at only one locality, but it is improbable that it has been transported t’here. The Glass is mixed with vein quartz gravel. The vein quartz is nearly 100 per cent SiO, and does not resemble the composition of the Darwin Glass. It is t,hus an improbable source material.

ISOTOPE DATA P;o argon of radiogenic origin wa,s detected, and this limits the a’ge of the glass to less than about one million years (Dr. I. McDouas~~, personal communication). The Sra7/Srs6 ratio was O-810 on sample 4F. The tot.al r~lbidiu~~l content, was 75 ppm and strontium, 15 ppm: measured by isotope dilution. It is not possible to calculate a significant age because the initial SP Sr8” ratio is unknown, and the history of the parent material is uncertain. It is also possible that some redist~but.ioll or loss of Rb and Sr has occurred during melting. X formal calculation, based on nn initial Srs7/Srs6 ratio of 0,710 gives an age of zi5O M 10” years. DARWIN GLASS The Glass is frothy and vesicular, and very irregular in shape, with twisted stalactitic forms. Numerous bubble pits are present and pieces are thus of low density (S.G. 1~7-2.2). The samples analysed weighed from 4.3 to liz g. Many of the pieces appear to be broken at the ends, with slight roulldi~g, but there is little evidence of abrasion, and the glass is easily fragmented, and could not survive st~ream transport for any distance. Twisted and bent particles of fused silica (lechatelierite) are present. Some carbonaceous matter is left, following digestion with HF and I&Sop. A representative collection of samples was made, and eight, fragments were selected for analysis. Figure 2 shows typical forms. Samples A, C. D, F, H., J, K and I, were analysed. ANALYTICAL METHODS S%npZe preparation.

The samples were carefulIy cleaned and washed in distiilecl u-atcr. They were then crushed by gentle tapping in a hardened steel mortar (composition : l”//o C’. 2% Mn, 0.3% Si). The glass was friable and easily reduced t,o pass 20 mesh silk bolting cloth. They were then ground in a mechanical agat,e mortar to pass 120 mesh silk bolting eloth mounted in plastic sieve holders. Many samples have been prepared by this procedure also used for the preparation of tektites (TAYLOR,1962). The method has been tested and there is no evidenre of contamination (TAYLOR,1962) by such critical elements as nickel. AEka-li elements. Sodinm and potassium were determined using a Perkin Elmer model 136

ud

II,

The geochemistry Table

2. Data

of Darwin

Glass

473

for UarM%n Glass samples

wt. so. 1. 2. 3. 4. 3. 6. 7. x.

A c: I) E’ H .I K L

(g)

S.G.*

K.1.

6.662

1.724 2.187 1.886 1.969 2.152 2.070 2.184 3,178

I.4810 1.4885 1.4795 1.4795 1.4790 1.4775 1.4790 1.4775

4.979 6.172 11.925 8.002 4.361 6.101 4.3”7 . d.

* S.G. determined

on whole samples.

frame photometer. The following spectrographic conditions were common to the determination of Li, Rb and Cs. A Hilger large glass spectrograph (E.744) was employed in the wavelength The arc was focused on the slit, using a Hilger range 48OU-11,500 A, with Kodak IN plates. F.S.18 lens. a seven step sector (2: 1 ratio) was used, with steps 6 and 7 set to transmit 50 per D.C. anode excitation was employed, with a cent so t,hat in effect, 8 steps were available. current of 4 A. Plates were developed in D.19.b developer for 4 min at 68”F, and fixed in Kodak rapid liquid fixer, using a Jarrell-Ash model 34-100 phot,oprocessor. The lines were read on a ,Jarrell-Ash model 23-100 densitometer losing a slit, width of 7 ,L and a drive rate of O..Smm/min. Kelati1.e intensities were obtained losing the self-calibration method (AHRETSS ,mtl l'au~ox, 1961, section 11-5) and background corrections were mada using a seidel function. Lithium rtn.rl IZzcbidiuwz. The unmixed poxvders were arced until the end of the alkali metal tlistillation period (AHHENS and TAYLOR, 1961, section 13-3). Anodes were Xa’at,ional Carbon Co. I)reformrrl electrodes Cat. So. L.3705 (YPK grade graphite) and cathocles were Sational Carbon (‘0. L.3863 &” carbon rod. The slit width was 20 11. Sodiklm was used as the internal standard and the following lines were read. Sa 5895, 5899, Li 6707 ant1 lib 7800. Values usecl for the st,a,ndartl samples are gi\,en in Table 3. Cesi2olL. The low alkali content of t.he samples necessitated addition of an alkali salt to ensure adeq~~ate sllppression of background to enable cesium to be determined. Tnoparts of sampleaere mixed \+?t,h one part of Johnson-Natthey Specpnre lithium carbonate, and arced unt,il the end of the alkali met,al distillation period, shown by t,he disappearance of the red lithium colour. (‘s 8521 was read, and the internal standard line K 6939. Anodes were Sational Carbon Co. SI’K grade C’at. T\‘o. L.4260. Volatile u7~d inrolatile groups. These elements u-eredetcr,ninrd;tccortlingto the genera.1 scheme l)roposetl by AHISENY and TAYLOR (1961, section 13-3). The following operating conditions A Jarrell&Ash Ebert grating spectrograph (model 71-100) were common to both procedures. was used \vith a 15000 line per inch grating in the tist order (dispersion 5.2 A/mm). The arc was focused on the slit with a 450 mm cylindrica,l quartz lens, and a seven step sect)or (2: 1 ratio) was used. Kodak 103-O plates were llsed. The developing conditions, plate calibration and densitomet,ry were identical to those described \mder the section on the alkali element,s. I Toldilr group. Two parts of t’he sample were mixed with one part of Johnson -Matthey Specpure lithirun carbonat’e, and arced at 5 i), d.c. (anode excitation) until the end of t,he alkali lnetal tlistillation period. Anodes were Kational Carbon Co. graphite preformrtl rlectjrodes <‘at. So. I~.4006 ant1 cathodes were Xational Carbon Co. -kin. carbon rod (Cat. Xo. L.3863). .I preexposure period of 5 set was 1~~1 to aroicl C’X emission in the initial stages of arcing. Slit width was 30 I’. The following lines were read? I’b 2833. 3683. 4057, Ga 2944, C!u 3274. -lg 3380. Sn 3175. Tl 3775 and Zn 3345 were not detected. \*al\~s of standards use(l are given in Table 3. SP-2gTaphite Indtrtil~ yroup. The samples were mixed with t\vo partsofNationalCarbonCo. powder. rontaining 0.004 per cent Johnson-Matthey Specpure (NH,),Pd(KO,), using a Spex model 5000 mixer mill and plastic \-ia,ls for mixing. Samples a-erc arced to completion. Anode

were Sat,ional Carbon Co. SPK gra,phit,e preformed electrodes Cat,. So. L.37O.j ant1 cat tlotlc%s were h’ational Carbon Co. & in. carbon rod. The slit, width was 16 p, and current ~va5 5 .I (l.c. Pd 3421 was used as the internal standard and the following lines were read. Mg 2780. Mn 2801. Fe 2937, Ti 3242, Cu 3274, Ni 3414, Zr 3438, Co 3453, Ca 3158, Ca 4226, SC 4246. C’r 12.?4, Y 4374, La 4333, V 4379, Sr 4607, Ba 4554, 4934. Background corrections SWIY ~neltltz on (‘0 3453. SC 4246 and \- 4279. Values used for standard samples are given in Table 3. Table

Ga Cr Fe % -Ilg % Li Ti Ni CO

Cu V Zr ?vln SC Y Ca % Sr Pb Ba Rb ck

3. \-alues used for standard

(t-1

TV- 1

18 15 1.37 0.24 26.5 1400

22 120 7.76 4.0 10 6500 70 50

-

10 13 185 210 17 0.97 250 50 1220 220 1.5

110 240 120 1300 34 40 120 8 165 22 -

samples*

and precisioni_ USBS

USBS

102

165

Sp-1 22 58 5.94

3000 40 19 20 98 2200

of spectrographic

7.4 0.46 0.13 3.4 950

methods

-

-

240 39 -0.006 20

200 300 345 1.3

24 5.4

* Data are in parts per million except where indicated in per cent. t Precision is expressed as relative deviation (C), or per cent standard

c 5 4 4 4 6 4 4 7 10 4 7 ;i Y

13

18 410

I’recision

-

i

6 Y 7 ;, 7 10

deviat,ion.

ilfajor elements. C’hemical analysis for SiO,, Al,O,, Fe,O,, FeO, MgO, TiO, ant1 H,O ( + and -) were carried out on five samples of Darwin Glass by Dr. H. B. R‘iik. Helsinki. Values for CaO were determined spectrographically by Mrs. A. Halsey in all sample. its wrrc FeO, MgO and TiO, in samples 2, 6 and 8. Determinations of SiO,, Al,O,, Fe,O,, FeO, P,O,, CO, and H,O ( $ and -) ww rarrircl out chemically by M. Chiba (Japan Inspection Co.) on the Tasmanian sandstones. Sprctrographic total iron values agree with t,hose determined chemically. Mrs. A. Ha.lsr>- tlcternlinccl MgO, CaO and TiO, spectrographically in these samples. The Na,O and I(,0 I-alur~ ill the sandstones and Darwin Gla,ss were determined by *T. Cooper nsing a Perkin Elmr~ >Inrl~l 146 flame photometer.

A~ALYTICAI,

RESULTS

The data for the Darwin Glass samples are given in Table 4 (major elements expressed as oxides) and Table 5 (element abundances in weight per cent or parts per million). Tables 9 and 10 record similar data for the Tasmanian sand&ones. The analytical precision, expressed as a relative deviation in per cent, is given in Table 3. Analytical experience in the last ten years has demonstrated the tlifliculty of assessing the accxracy of data (FAIRBAIRN et ul., 1951. The spectrographic

The geochemistry

477

of Darwin Glass

Tahle 4. Chc,mica,lcomposition of Darwin Glass, expressed as oxides (wt. per cent) II

I

__-

1

2

3

4

.I

c'

J)

F

SiOf _U,O, Fe,O, Fe0 XgO (‘a0 se,0 K,O TiO, H,O+ H,O-

84.08 7.45 2.65 1.60 1.24 0.12 0.065 1.99 0.58 0.18 0.01 99.96

E’eo*

3.!M

84.94 1.36 0.19 oa75 1.77 0.m -

2.29 1.60 0.93 0.052 0.047 I.96 0.58 0.14 0.01 ‘9‘3 i ‘I _

85.20 6.55 2.41 2.10 0.94 0.088 0.044 1.78 0.56 0.15 0.0.5 99.89

3.6.5

4.27

i.17

3.30

(i J

5 H 86.84

~-

6.25 2.10 1.60 0.87 0.11

0.68 0.10

0.035 1.78 0.56 0.03 0.06 100.24 3.49

0.042 1.66 0.59 2.07

i

x

K

L

87.05 5.80 2.23 140 0.98 0.080 0.050 1.59 o.:i5 0.0s 0.00 100.03 3.41

II av.

O\Y?l%ll it\:.

86.01 6.44 2.26 1.68 0.84 0.085

8.i.62 6.64 2.34 1.66 0.96 0~10

0.042 1.72 0.63 -

0,044 1.78 0.58 0.10 0.03 99.85

0.051 1.81 0.59 0.12 0.03 99.92

2.12

3.17

3.31

-_ 0.66 0.078

* Total iron expressed as FeO. -: not determined. ;Inc~Z@n. H. B. Wiik ;\. Halsey: CEO throughout. FeO, MgO + TiO, in 2, G and 8. J. Cooper: (Sa,O and K,O.).

methods were calibrated with available standard samples of similar matrix, wherever possible. The t.wo rock standards, G-l (granite) and m7-1 (diabase) (AHRENS and FLEISCHER, 1960), the Canadian syenite No. 1 (WEBBER, 1961), the U.S. Bureau of Standards No. 102 (silica brick) and 165 (glass sand) were employed as calibration standards, with the values shown in Table 3. The accuracy of the analytical results is directly related to the values used for these standards. COMPOSITIONAL

DIFFERENCES

AMONG

DARWIN

GLASS

SAMPLES

The range in silica composition is quite restricted; it varies from 84 to 8i per cent,. MgO, CaO, Na,O and total iron vary by a factor of about, two; the other major constituents have a lesser spread in composition. The trace elements sho\J wider variation. The majority of the elements (Ga, Li, Ni, Cu, Zr, SC, Y and Pb) have a range in concentration of about a factor of two. Cr, Mn and Co vary by larger factors and only V, Sr, Ba and Cs vary by factors of less t’han two. The range in composition is given in Table 5. In Tables 4 and 5 the analyses are divided into two distinct groups, labelled 1 and II. Averages of both groups are given, in addition to the overall average. The ratio between the two averages is given. Silicon is lower in Group I and all ot,her elements determined are present in greater amount than in Group II. Possibly, the grouping is partly due to the small number of samples. However, several elements are present in Group I in markedly greater abundance. Those which are most significant are, in order Co ( x lo), Mn ( x 2-58 in Group I). Cr ( x 2.07). (‘a ( x 1.92), Ni ( x l-85), Y ( x 1*58), Na ( x 1.58), Cu ( x l-57), Mg ( x 1.55). Rb ( x 1.49).

I,i ( >: 1.48).

I-

‘&;!I

The geochemistry of’ lIarwin Glass

It is significant that both Ni and Co are much higher in Group I. If the grcat’et abundance of these elements is due to a meteoritic contribution, this might be responsible for the lower silica value, but could scarcely account for t,he increased Most of these would not be supplied in quantity amounts of the other elements. cxven from a stony meteorite. It is curious that the two samples which contain the best evidence for a meteoritic contribution, possess a greater conceutration of so many other constituents, notably chromium. An analogy with the composition of t,he billitonite group of tektites is suggested by this dat’a. COIVPARISONWITH TERRESTRIAL ROCKS The assumption is made here that the present, composition is little modified from bhat of the parent material from which Darwin Glass is derived. Evidence to support this view n-ill be considered in the discussion on the data for the alkali elements. (a) Igneous rocks. In WASHINGTON’S ( 1917) compilation of 5159 superior analyses of igneous rocks, only one sample has a silica content within the range of t’hat of Darwin Glass (84-90 % SiO,). The analysis, (SiO,, 84.15 % ; Al,O,, 9.67 7: : Fe,O,, 0.51%;

FeO,

0.07 %;

MgO,

0.04%;

CaO, O-53”/:;

Na,O,

2.65%;

K,O,

1.57 %;

Ti02, trace) is labelled “beresite”. It bears no resemblance to the composition of Darwin Glass, the notable exceptions being iron: calcium, magnesium and sodium. In general, the high concentrations of silicon, iron, magnesium and the low values for sodium indicate that Darwin Glass is not a product of terrestrial igneous processes. (b) Xedimentary rocks. The high silica content of Darwin Glass invites comparison with the silica rich sedimentary rocks. Two recent compilations of major element data in sandstones are available (MIDDLETON, 1960; PETTIJOHN, 1963). Both sets of averages are close. Those of M~DDLETON (1960) which provided geometric means and were recalculated to 100 per cent for SiO,, Al,O,, total iron as Fe,O,, MgO, CaO, Na,O and K,O, (excluding H,O, CO,, loss on ignition, etc.), are used (Table 6) in this paper. The average silica contents of subgreywackes (78.7 per cent). arkoses (82.8 per cent) and quartzites (93.3 per cent) bracket the silica contenh of Darwin Glass. The iron and magnesium contents of arkose are too low to make a valid comparison. A 1: 1 mix of quartzite and subgrepwacke produces a composition rat,her close to that of Darwin Glass (Table 6). Apart from the silica content, the most unusual feature in the major element abundances is the nearly total lack of calcium and sodium. The lat’ter element could be lost in a selective distillation process during melting, but calcium will not be depleted by this process. Potassium, which would be lost more readily than sodium is present at about two per cent and the K/Na ratio cannot be explained reasonably by selective loss of sodium. It is preferable Do assign the lack of sodium and calcium to an absence of plagioclase feldspar in the parent material. The low sodium content is also diagnostic in limiting the amount of K-feldspar, which commonly has a K/Na ratio of about 5 (HEIER and TAYLOR, 1959). On this basis, a sodium content of 0.05 ‘A gives a maximum potassium contribution of about 0.3 % K from K-feldspar. The modal K-feldspar would be about 2-3 “/o. Most of the potassium (1.5 %) would be contributed from other sources. The relative concentrations of K,O (1.5 %), MgO (about 1 %), total iron (3.3 “/o FeO) and TiO, (0.6 04,) are consistent with the composition of an illite type clay. The uncert,aint#y in illite

S. R. TAYLOR and 31. S~WMOS

480

Table 6. Comparison of’ average sandstone analyses with Darwin Glass (wt. per cent)

SiO, Al&, Fe0 MgO cao Na,O K,O

-___

1

2

3

4

78.70 11.77 3.20 1.23 2.39 0.84 1.51

82.75 10.58 1.75 0.23 0.90 0.71 2.88

93.34 3.33 1.34 0.35 0.36 0.31 0.82

86.0 7.55 2.27 0.79 1.33 0,.58 1.17

.i

85.6 6.64 3.31 0.98 0.10 0.05 1.81

1. Average subgreywacke (geometric mean) (MIYULETOX, 1960). 2. Average arkose (geometric mean) (MIDDLETON, 1960).

3. Average orthoquartzite (geometricmean) (MIDDLETON, 1960). 4. 1:l mix of columns 1 and 3. 5. Average Darwin Glass (thispaper).

The iron concentration is composition makes any formal calculation difficult. probably too high to be derived entirely from such a source, and about one per cent iron oxide would be needed to balance the calculation. The nickel concentration, if derived from meteoritic sources, indicates that the maximum iron contribution from such a source would be about 0.25 %. In summary, this comparison of the major element composition with possible terrestrial source materials suggests that an argillaceous sandstone, with about 80 % quartz, 15 “/o illite type clay, 2-3 % K-feldspar and 1 % iron oxide, is the closest match. This composition lies about on the boundary between quartzites and subgreywaclres. It may be noted that the composition of australites is close to t,he subgreyu-ackeeshale boundary (TAYLOR, 1962). COMPARISON

0F TRacE

IN SANDSTONES il compilation

~XLEIMENT Axxn4Nms AND

DARWIN

G~sss

of sandstone analyses has recently been made by PETTIJOHS (1963). Adequate data for the various sandstone classes, except perhaps greywacke, are not yet available. The difficulties of classification of sandstones, the varying conbents of quartz, feldspars, clay, cement and heavy minerals, and the lack of geochemical incentive to investigate sandstones, have resulted in little comparative data being available. PETTIJOHN (1963) comments that “one can sometimes make better estimates of abundance of an element in sandstone based on its known geochemical behaviour than . . . by simply averaging the concentrations found by a feu sporadic or poorly documented analyses” (p. S.11). In addit’ion to the data summarized by PETTIJONK (1963), TUREKIAN and WEDEPOHL (1961) have provided averages for trace elements in sandstones. Both sets of data are given in Table 7 together with the Darwin Glass average. The data on greywackes from ~MACPHERSON (1958) and in particular from WEBER anal MIDDLETON (1961) are useful in the present context for discussing the distribution of elements among the various fract,ions (quartz, clay, feldspar, detrital minerals) but are less directly applicable to the more silica-rich samples studied here. Data from other sources are discussed in the text. The elements are given in the conventional geochemical order of increasing ionic radius. Although this is not a significant

The geocbenlistry Titble

1

2

SC x7

s;r Pb Ba Rb cs

12 3.5 1.5 1500 2 0.3 s 20 220 SO

2 Average

3

-_

.3.6&10 liSpSOT, 2.&;,.1

so--3480 -

1500 2

3360-47X0 82~205 <3-27 6.4-16 23 -3s 220-490 82K300 2..5 ~5.2

I

40 20 7 X0 60 0.X

26-43 170-450

4 35 9 300 60

anti

Hange

s-10 10~20 16

37-820 140~-77.-,

481

Glass

7. ‘Trace element cont,ent of santlstones lhwin Glass (dRt,a in ppm)

IIangtt c&l Cr Li Ti Ni Co CL1 I_ Zr Mn

of I)Mv~I~

11 -26 13&16 4,tiGll 290 -360 61-110 3’2 4.3

8.4 113 3.7 3840 122 10 29 390 1% 4.1 18 14 7.6 340 78 3.6

1. TUREKIAN and ~VEDEPOHL(1961). 2. PETTIJOHN(1963). 3. Darwin Glass (this paper). X : order of magnit1tde. -: no data given.

factor for the present distribution of elements in Darwin Glass, it was no doubt an important factor in controlling the abundances in the source material. Gulliuna BURTC)N et al. ( 1959) give a wide range of gallium values in various sandstones from 0.8 to 15 ppm. PETTIJOHN (1963) estimates that the Ga content of most sandstones is within the range 5-10 ppm, and TUREKIAN and WEDEPOHL (1961) estimate an average of 12 ppm. The Darwin Glass data vary from 5.6 to 10 ppm and average 8.4 ppm. This gives an Al/Ga ratio of 4220. The crustal average (Al, 8.13 per cent ; Ga, 19 ppm) is 4280. If Darwin Glass is terrestrial in origin, the Al/Ga ratio has not been significantly altered during melting. This ratio, and the absolute abundance of gallium, is consistent with a sandstone type parent material. WEBER and MIDDLETON (1961) found that all of the gallium was contributed by the clay fraction in greywackes.

C’hromium The distribution of chromium in sandstones is erratic, and a wide range of values has been reported. SAHAMA (1945) reported 68-200 ppm in quartzites from southern Lapland. Frijhlich (1960) gives values ranging from 1 to 150 ppm with an average value of 59 ppm (excluding greywackes). SHOEMAKER et al. (1959) give an average of 7 ppm for 289 samples of the Colorado Plateau sandstones (dominantly arkoses and orthoquartzit’es). The chromium content Tvill depend par+ on the presence of

detrital chromit,e, and on the abundance of illite (FR~HLICH, 1960). X sandstone with 20 per cent illite could contain about 40-50 ppm Cr supplied by the clay. _%rgillaceous sandstones should therefore contain more Cr than arkoses. %7~~~~ and MIDDLETON (1961) found that the clay fraction of greywacke contributed about half to two-thirds of the chromium, with heavy minerals making a substantial contribution. These two sources of chromium in sandstones no doubt account for the wide range in reported values. PETTIJOHN(1963) estimates that the true average Cr content of sandstones is in the range lo-20 ppm and for orthoquartzites and arkoses, this statement is very reasonable. The average Cr content of Darwin Glass is 115 ppm (Group I : 190 ppm ; Group II: 92 ppm) and such concentrations are consistent with an illite bearing sandstone with detrital accessory chromite. The high chromium content of Group I is noteworthy. Lithium STROCK (1936) gives an average value of 17 ppm for sandstones. HORSTJLAN’S (1957) average is 12 ppm if a glauconite sandstone with 25 ppm is excluded from the average. Horstman estimates the lithium content of illite at about 40 ppm (1~. 10) and about 20 per cent illite should contribute 8 ppm Li. The Darwin Glass range from 2.8 to 5.1 ppm is a little low, but within the uncertainties of the calcula.tion. Titanium The titanium average for Darwin Glass (3540 ppm) is higher than the sandstone average of 1500 ppm given by TUREKIAN and WEDEPOHL (1961) and PETTIJOHN (1963) but is close to the average (3470 ppm) of 158 analyses (3709 samples) of Russian platform sandstones (VINOORADOV and RONOV, 1956). WEBER and M~DDLETOR(1961) found that about 10 per cent of the total titanium content could be contributed by the heavy mineral fraction, with the rest in the clay fraction. The high Ti values in Darwin Glass are consistent with the high concentrations of iron and magnesium.

The nickel content of sandstones is in general very low. Both PETTIJOHN (1963) and TUREKIAN and WEDEPOHL (1961) give averages of 2 ppm. SHOEMAKERet al. (1959) found the same average value for 289 samples of Colorado Plateau sandstones. Only the greywackes have higher nickel contents. WEBER and MIDDLETON (1961) found averages of 27 ppm Ni in arkosic greywackes and 43 ppm in greywackes. WEDEPOHL (1963) reports an average of 22 ppm Ni in a composite sample of 23 feldspathic Buntsandsteine. MACPHERSON(1958) found up to 90 ppm Ni in greywackes. The nickel content of Darwin Glass (Group I, 185 ppm ; Group II, 100 ppm) is an outstanding anomaly in comparison with the comparabel sandstone averages of a few parts per million. From the data of WEBER and MIDDLETON (1961) the Cr/Ni ratio in the heavy mineral fraction and the clay fraction is about three. Accordingly, about one-third of the nickel content of Darwin Glass (60 ppm in Group I and 30 ppm in Group II) might be original. The remaining two-thirds at

The geochemistry

least are anomalous for terrestrial considered in the next section.

of Dar\vin Glass

sources.

The evidence

453

from the Ni/Co ratio is

Cobalt and Ni/Co ratios There are few data for cobalt in sandstones. TGREKUN and WEUEPHOL (1961) Give an average value of 0.3 ppm for quartzites. CARR and TCREKIAE (1961) give a Value of 2-T ppm Co in sandstones, and ISIIIBASHI (1959) report’s an average of -4ppm. The Colorado Plateau sandstones contain 1 ppm Co (SHOE~~AKER et al., 1958). The Group II Darwin Glass samples contain cobalt nearly at the detection limit’ of t’he spect’rographic method used and is estimated at less than 3 ppm. (Cobalt was not detected in G-l at 2.3 ppm.) The cobalt value in Group II is thus consistent with that to be expected in sandstones (greywacke types contain 15-22 ppm according to WESER and MIDDLETON, 1961). The low cobalt content in the Group 11 Jlarwin Glass samples is also significant in placing an upper limit on the nickel content’. The average terrestrial Ni/Co ratio is about two. If the source material of Darwin Glass is terrestrial, this cobalt content limits the terrestrial nickel to less than 10 ppm, and estimates based on Cr/Ni ratios must be revised downwards. This means that about 90 per cent of the nickel in Group II, or 90 ppm, is anomalous. The Ni/Co ratio in iron meteorites averages about 15, so that only a few ppm Co would be contributed from such a source. The cobalt content of the Grollp I Darwin (i-la,ss samples (27 ppm) would .be consistent with about 50 ppm nickel if derived terrestrially. This accords with the 60 ppm Ni deduced from Cr/Ni ratios. Thus 60-90 per cent of t,he nickel in Darwin Glass is anomalous, if derived from a terrestrial source.

PETTIJOHN (1963) gives lo-20 ppm as the most reasonable values for copper in sandstones. HARTMAN (1963) gives values between 3 and 10 ppm for Permian and Triassic sandstones from Germany. The average for copper in the Colorado Plateau sandstones is 9 ppm (SHOEMAKER et al., 1959). WEBER and MIDDLETON (1961) report 9 ppm Cu in arkosic greywackes and 33 ppm in greywackes. MACPHERSON (I 958) gives an average of 48 ppm Cu for greywackes. The Darwin Glass range from 6.4 to 16 ppm, with an average of 10 ppm Cu is consistent with the sandstone averages quoted. There is no evidence to suggest depletion, or loss of copper during t,he melting process.

The Colorado Plateau sandstones (SHOEMAKER et al., 1959) average 11 ppm T’. The content, of vanadium is a good index of the clay content of sandstones, although some will also be contributed from heavy minerals, according to the data of WEBER and &IIDI~LETOK(1961). The vanadium averages reported by the latter workers for arkosic greywackes and greywackes are 43 and 67 ppm respectively. WEDEPOHL ( 1963) reports 19 ppm V in a composite sample of 23 feldspathic Buntsandsteine. l’he Darwin Glass average of 29 ppm (range 23-35 ppm) is slightly higher than the average of 20 ppm suggested by TUREKIAN and WEDEPOHL (1961) or the lo-20 ppm of PETTIJOHN (1963). This higher value is in agreement with t#he high content of

4x1

S.R. TAYLOR

and 11.

So~oaror

iron. magnesium and titanium among the major elements. It is possible that further work on sandstones will show the value of elements such as vanadium for purposes of classification. Zirconium

This element is ubiquitous in sandstones due to the persistence of the resistant mineral zircon during weathering and transport. DEGENHARDT (1957), in his extensive study of the geochemistry of zirconium, gives an average of 220 ppm for sandstones, with a range in composition from 33 to 570 ppm. TUREIUAK and WEDEPOHL ( 1961) give an average of 220 ppm, and PETTIJOHN(1963) gives an average of 200-250 ppm with a range from 37 to 820 ppm Zr. SHOEMAKER et al. (1959) report an average of 88 ppm for the Colorado Plateau sandstones. WEDEPOHL (1963) gives an average of 370 ppm Zr in a composite sample of 23 feldspathic Buntsandsteine. The Darwin Glass average of 390 ppm is well within the range of Zr values reported in sandstones, and the range in composition (220-490 ppm) is bracketed by the range of zirconium values in sandstones. The high Zr figures are contribubory evidence for the operation of weathering processes during formation of the source material of Darwin Glass. Manganese

The distribution of manganese in sandstones is variable due to its presence in cementing material (WEBER and M~DDLETON, 1961; HARTMAN, 1963). The Colorado Plateau sandstones average 140 ppm Mn, very close to the Darwin Glass average of 135 ppm. WEDEPOHL (1963) gives an average of 170 ppm Mn in a composite sample of 23 feldspathic Buntsandsteine. WEBER and MIDDLETON (1961) found that about equal quantities of Mn were supplied by the cement and by the heavy minerals. The low content of Mn, coupled with the low abundance of Ca and Sr is consistent, with a source material containing little calcareous cement. Possibly many of the chemical manganese determinations are in error at low concentrations, and spectrographic determinations are superior at this level (AHRENS and TAYLOR, 1961).

Very few data are available. TUREKIAN and WEDEPOHL (1961) estimate an average of 1 ppm SC in sandstones. The Chinle formation sandstones of the Colorado Plateau contain 4 ppm (SHOEMAKER et al.,1959). The arkosic greywackes studied by WEBER and MIDDLETON (1961) contain 5 ppm SC and the greywackes, 8 ppm. The Darwin Glass range in composition from 2.5 t’o 5.2 (average 4.1 ppm) is consistent with a sandstone type source material.

Data are again very sparse. TUREKIAN and WEDEPOHL (1961) give an average of Y. The average of the Colorado Plateau sandstones is 4 ppm am1 the Chinle formation, 16 ppm Y. The content in Darwin Glass varies from 11 to 26 ppm, and averages 18 ppm. Thus the Se/Y rat,io is close to that in the Chinle sandstones. 40 ppm

The geochemistry

of’ Darwin Glass

-cx.s

Strontium PETTIJOHN (1963) gives an average of 35 ppm Sr in sandstones, and TLXEKIAN The Colorado Plateau sandstones average and WEDEPOHL (1961) give 20 ppm. 45 ppm (SHOEMAKERet al., 1959). WEDEPOHL (1963) gives an average of 78 ppm Sr in a composite sample of 23 feldspathic Buntsandsteinc. The low value of Sr in Darwin Glass (14 ppm) reflects the low abundance of calcium.

WEI)EPOHL (1956) gives 7 ppm lead as t,he average in sandstones and TURENA-? and WEDEPOHL (1961) quote this figure. PETTIJOHN (1963) gives a value of 9 ppm_ HARTMANN ( 1963) found that lead was in general between 2 and 10 ppm in German Darwin Glass contains 4.6 to 11 ppm, w&h an Permian and Triassic sandstones. average of 7.6 ppm, very close to t’he sandstone dat)a. There is no evidence of selective loss of lead from Darwin Glass in comparison with assumed parent materials. Bnrium PETTIJOHN (1963) gives an average of 300 ppm Ba in sandstones, with a range The Colorado Plateau sandstones cont,ain an average of from 170 to 450 ppm. 280 ppm (SHOE~\IAKER et al., 1959). WEDEPOHL (1963) gives an average of 620 ppm WEBER and MIDDLETOX Ba in a composite sample of 23 feldspathic Buntsandsteine. (1961) report averages of 220 ppm Ba for arkosic greywackes and 380 ppm for greywackes. The Darwin Glass average of 340 ppm and range from 290-360 ppm is consistent with the sandstone data. Most of t’he barium will accompany potassium in the clay minerals (WEBER and MTDDLETON,1961). Rubidium Three non-glauconitic sandstones analysed by HORST~IZAX(1957) gave values of 20, 50 and 80 ppm Rb. The inclusion of a glauconitic sandst’one with 100 ppm Rb raises the average to 60 ppm, the value quoted by TUREKIAN and WEDEPOHL (1961) and PETTIJOHN( 1963). WEDEPOHL (1963) found an average of 85 ppm in a composite sample of 23 Buntsandsteine (K/Rb = 260). In well mixed sedimentary rocks, the K/Rb ratio may be expected to be about the crustal average of 220, although Rb (and Cs) might increase relative to K in clays. The Darwin Glass samples range in concenCration from 61 to 110 ppm, with an average of 78 ppm. The average K;Rb ratio is 192. The Group I K/Rb ratio is 150. In Group II, the ratio varies from 181 to 224, and averages 211, near the crustal average. This is an interesting and significant difference between the two groups, and would be consistent with a greater clay content in the Group I source material. This is also indicated by t,he concentrations of iron, magnesium, etc. Rb would be lost more readily than K in any selective distillation process during melting. The low K/Rb ratios in Group I provide. strong evidence against the operation of such a process and the K/Rb ratios in Group II, which are in general somewhat less than the crust’al average; likewise indicate no change in the relative proportions of K and Rb during melting. 6

C’esiurn There are very few data on the concentration of cesium in sandstones. HURYTMBN (1957) gives a value of 9 ppm for a glauconitic sandstone, the only sample in which the element was detected. Darwin Glass averages 3.6 ppm Cs, with a K/Cs ratio of about 4000. (It was not possible to determine cesium in all samples because of shortage of material, and a proper comparison between Groups I and II cannot be made.) This compares with a crustal K/Cs ratio of about 7000. This evidence is important for two conclusions : (a) a high cesium bearing material (e.g. clay) was present in t’he source material, (b) the low KiCs ratio is evidence against selective loss of cesium during melting. If the readily volatile cesium shows no evidence of selective loss, we may assume that such a process was not’ important during the melting of Darwin Glass, and we are just’ified in the assumption that the present composition is near to that of the source material, for the elements considered here.

The following trace elements are present in Darwin Glass in concentrations similar to those in sandstones. Ga, Ti, Co (Group II). Cu, Zr, SC, Y. Pb: Ba, Rb (Group II). The elements Li, Mn and Sr are lower than the average concentrations in sandstones. The low content of Sr is consistent with the low Ca value. Together with the low concentration of Mn, it would be consistent with a lack of cementing mat,erial. Cr, Co (Group I): V, Rb (Group I) and Cs are present in greater abundance than typical of sandstones. Of these elements, the concent.rations of V, Rb, Cs and possibly Co would be those predicted in an argillaceous sandstone. Some of the Cr could be attributed to such a cause, but the high concentration requires a source such as chromite or chrome spinel, in a terrestrial rock. The Al/Ga and Se/k’ ratios are the same as the crustal average. The low KiRb ratios in Group I, and the low K/Cs ratios would be consistent with the ratios in an argillaceous clay, and indicate no selective loss of Cs or Rb relative to K during melting. The composition is accordingly probably close to that of the source material for the elements considered here. The concentration of nickel is anomalously high in comparison with terrestrial rocks of analogous composition. COMPARISON WITH ADJACENT COUNTRY ROCKS In the section on the geological occurrence of Darwin Glass, it was noted that sandstones occur underlying part of the area. Samples of the two principal formations, the Owen Sandstone and the Crotty Sandstone, were analysed, and in addition, two samples of soil and gravel. Darwin Glass has sometimes been confused with slag from a nearby copper smelter, and a sample of this material was also analysed. Locality data for these samples are given in Table 8, and the chemical analyses in Table 9 (major elements expressed as oxides) and in Table 10 (elemental abundances). The mineral compositions of the Crotty Sandstone and Owen Sandstone are given in the section on geology. Both the Owen Sandstone and the Crotty Sandstone are orthoquartzites (PETTIJOHN, 1957). They are very high iii Rio, (>M per cent) and accordingly IOU. in

Sample nllrn brr

Locality

Desrript ion Owen Sandstone-Conglomerate gravel Dark grey clay and gravel O-l ft below surface. Pieces of Darwin glass were removed Crotty Sandstone

9.

klL.358

l/4 mile souhh of Darwin

10.

>IL.359

Stl.mo as So.

11.

NL.360

Andrew

12.

ML.228

13.

Table

FCO”

9

Divide,

1 mile

sout’h of Crotty Ten Mile Hill. 60’ below summit, east side Crott~

Soil and gral-cl mixture, O--6 in, dept.h Slag from smekcr

9. Chemical composition of Tasmanian s:mdstones, expressed its oxides (wt’. per cent,) 9 53.L.368

10 ML.359

98.72 0.14 0.16 0.02 0.20 0,004 0.0065 0,013 0.13 0.22

98.84 o-13 0.13 nd 0.15 0.004 0.0073 0.010 0,067 0.13 0.12

0.10 0.03 0.12 99.86 0.17

0.02 0.14 99.75 0.12

11 ML.360

12 ML.228

95.6i 1.87 0.17 0.06

-

0.23 0.021

0.20 0.020

0.011 0.61 0.22 0.60 0.16 0.05 ml 99.67

0.018 0.040 0.18 -_. -

0.22

0.13

* Total %‘c!esprt:sstYl as FeO. : not de&mined. Artulysts. M. Chiba. MgO, CaO, TiOz: A. H&e)-. X’:t.$. 9. ML358 Owen Sandstone gravel. 10. ML359 gravel overlying Owen Sandstone. 11. ML360 Crotty Sandstone. 12. ML228 Soil and gravel mixture from Ten Mile Hill.

K,O:

J. Cooper.

S. K. Taul,oR

488 Table

and M. SoLoM0x

10. Analyses of Tasmaniarl sandstones (9--l_“) and C’rott)y Slag lhta expressed as ppm except where indicated in per cent 11 JIL.:Gx) 90 46.2 0.07 2 100 0.09 0.09 1.o 400 11tl rltl 18 nd Ml 0.09 1x0 Jltt Jltl nti

54 :jo r1cl 6.1 0~008 nd ntl nd: not, detected. -: not determined. ilncrlysts: A. Halsey.

220 44.7 0.99 7 20 0.12 0.14 0.4 1300 Jl(i IJ(i 6

ntl 0.0.: (I.17 23-I fi nd ntl 78 150 nd 11 0.51 270 24

I’, Si. Al, Fc in 9, 10, 11: 31. (‘hi&~. Sit, K:

12 ML.228

1 BO 0.1” I.2 10.W ML 11tl 7 Id 0.10 125 nd JJCl IId

(13). 13

2x 3otr O.fi.5 XO(l

4x 43 2.5

1s.o 100 4300 4 r1tL

130 140 nd 5.4 0~033 100 nd

l-10 fi700

.J. Cooper.

otther constituents. In comparison with other sandstone data they are notably low in CaO. The determinations were made by spectrographic analysis, which is superior in accuracy to chemical methods at low concentrations (AHRENS and TAYLOR: 1961, Chap. 26). St’rontium was not detected (detection limit about 3 ppm) and this is an independent check on the low calcium value. The sodium and potassium values are low also. The Cr values are of interest. The maximum value for Cr is 315 ppm. and t,his is due to the presence of chromite, but this has not resulted in the addition of T’. Ni or Co. From the analyses of chromites and chrome spinels by Ross et al. (19fd), typical values are as follows: Ni. 1000 ppln; Co, 100 ppm; V: 100 ppm with Cr varying from i-5 to 38 per cent. Thus the presence of chromite in sandstones should not notably increase the content of Ni, Co or 1’. In the Owen Sandstone, the concentrations of SiO,. L&O,, MgO? CaO, Ea,O and K,O among the major constituents, and Ga. 12. Ti, V, Mn, Y, Sr, Ba and Rb exclude them as source material for Darwin Glass. CaO and Na,O, although low in Darwin Glass, are an order of magnitude lower in the Omen orthoquartzite. The only

The geochemistry

of &ruin

Glaw

489

const,ituent~s present in comparable amounts are chromium, zirconium, copper and lead. The Crotty Sandstone has comparable amounts of Ga, Cu, Zr, Pb and Ba, but t,hc concentrations of the other constituents are so different as to exclude this It is possible that an sandstone from consideration also, as source material. argillaceous facies of t,he Crotty Sandstone could be a suitable source. The differences in composition between these orthoquartzites and Darwin Glass cannot be reconciled by deriving the latter by a process of selective melting or distillation. The analysis of the slag (No. 13, Table 10) was carried out spectrographically. The composition bears no resemblance to Darwin Glass, but, is included here for coml)leteness. CO~\IPARISON U’ITH IMPACT GLASSES -1 comparison of the major element composition of impact glasses and .Darwin Glass has been given by TAYLOR and SACHS (1964, Table 8). With the exception of the black glass from Wabar: the SiO, contents are uniformly high. There is an overall similarity in the several analyses, marked by high absolute MgO content (for material with more than 85 per cent SiO,), low CaO and high K,O/Na,O ratios. The latter argue against much selective loss of alkalies during melting, in which potassium would be lost relative to sodium. The difference between the black and nhitc \l’abar glasses is mainly in the iron content, partly contributed from meteoritic sources. Several questions are suggested by these data. ( 1) 1s the similarit’y in composition coincidental? (2) Do impact glasses of distinct,ly different composition exist? (3) Is this general composition favoured for glass formation during impact 1 (1) These low alkali, high MgO and SiO, glasses show analogies with tektite composition. Is this comparison significant 2 There is clearly a need for an investigation of the geochemistry of the impact glasses, and work is planned on these lines. Modern trace element determinations for the impact glasses are restricted to nickel (lX:HnrAs~: 1960, 1962). The iron and nickel contents, and Fe/Ni ratios for meteorites. rocks, australites, impact glasses and Darwin Glass, are given in Table 11. The data are shown graphically in Fig. 3. The low Fe/Ni ratios of the impact glasses are notable for terrestrial material, and are only equalled by ultrabasic rocks. The Darwin Glass ratios are higher, but of the same order of magnitude as those for the impact glasses. and are distinctly different from the terrestria’l rocks and australit8es. COMPARISOK WITH AUSTRALITES At se\-era1points in the preceding discussion, similarities and differences between the composition of tektites and Darwin Glass have been noted. ,%dequate trace element data are available on1.v for the australites (TAYLOR, 1962; TAYLOR and SACHS; 1964) and discussion is restricted to a comparison with these tektites. The overall similarities of composition of the several tektite groups enable the conclusions to be extended to tektites in general. The australites are, in addition, the closest, tektit’e group; and: if t,here is ally genetic connexion with Darwin Glass,

S. K. TAYLOR

490 Table

and M. SoLozIos

11. lron and nickel average \wlnes for llletcorites, arlstralites, impact glasses and Darwin (:lnss

Reference

~. Iron meteorites Chondrites Ultrabasic rocks Basalt Granodiorite Syenite Granite Average igneolw rock Shale Greywncke Quart,zite Arwtralites Wabar Impact) Henbury glasses Aonelloul I Aouelloul Darwin Glass I II 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

1 2 3 3 3 3 3 4 .5 6 3 .> 7 7 x 9 10 10

Fe

Si

(76)

(PPm)

90.8 25.1 9.43 8.65 2.96 3.67 1.42 5 0 5.22 S-44 0.98 3.54 q5.42 4.42 1.40 2.15 2.91 2.46

85900 13400 2000 130 15 4 4.5

3.5 68 43 2 30 l’i0 520 “60 170 1x5 100

rocks,

Fe/Ni (M.) 10.6 18.7 47 665 1970 9180 3 160 1430 ii0 800 4900 1180 43 8.5 54 126 157 “50

fixASON(1962). UREY and CRAIG (1953). TCREKIAN and WEDEPOHL (1961). AHRENS and TAYLOR (1961). TAYLOR and RACKS (1964). WEBER and MIDDLETON(~~~~). EHMAKN (1962). EIWANN (1960). SXZITHand HEY (1952). This paper.

it would be expected to be displayed best by this group. A table and discussion of the relative composition of australites and Darwin Glass has recently been given (TAYLOR and SACHS, 1964, Table 7) and only the salient points are discussed here. The overall similarities are high magnesium contents and similar iron and potassium concentrations. If allowance is made for the differences in silica content (australite average is 73.3 per cent SiO,) then the concentrations of these element’s, and aluminium, are very similar. This similarity does not extend to calcium, sodium, and many of the tr.ace elements. Barium and sca’ndium are the only trace elements with comparable concentrations. However, most of t’he differences are explicable on the basis of different source materials. The Darwin Glass composition is similar to that of an argillaceous sandstone intermediate between subgreywacke and quartzite. Australites, on the other hand, resemble the composition of a sandy shale, intermediat,e between subgreywacke and shale (TAYLOR, 1962). Darwin Glass and australites are thus sufficiently different chemically to remove the possibility of a direct relationship. The australit,es more closely match average terrestrial element abundances than does the Darwin Glass.

The geochemidry

of Darwin

Glass

10%-

SIDERIlE

CHONORITE

0

0 /

1.0% -

c

ULTRABASIC

/ 1000

0

0

WARAR

-

0

HENBURY

NlCKEL

AOUE

2 LLOUL

/ -0

DARWIN

0

GLASS

0

l

l 100

a

CREYWACKE

C.P.M.

BASALT

/

-

0

0’

/

OAVERAGE

IGNEOUS

ROCK

AUSTRALIYE

/ 6 GRANODIORITE

10 -

/ 0

PER

CENT

SYENITE

IRON

Fig. 3. The relationship between iron and nickel for Lktrwin C&~SH, meteorit,es, impact gl~~ssw, alistralites nnd common t,ei-rest,rial rocks (&tit from Table 11).

492

S. R. TAYLORand 31. SOLOM~X

The most outstanding difference is the high absolute nickel content, and the low Fr/Ni ratios in Darwin Glass, compared to australites (Table 11 and Fig. 3). EVIDENCE FOR TERRESTRIAL OR EXTRA-TERRESTRIAL OKIGIX The evidence for a terrestrial origin may be summarized as fallows: (1) The overall composition is similar to that of an argillaceous sandstonrl (txscel)t for the nickel content and the Fe/Ni ratios). Such a composition implies that the source material originated through processes analogous to the complex series of events which result in the production of an argillaceous sandstone from prirnit,ivc solar system material. These are restricted to the surfaces of planets \vith tclrrcstrialike environments. (2) The friable, frothy nature of the glass and the absence of forms 1)roclucetl 1)~ The atmosl)hrric atmospheric flight (cf. tektites) argue for a local derivat,ion. break-up of a larger body could conceivably produce such fragments (1)~~analogy with the friable carbonaceous chondrites). The lack of observed meteorites with tile composition of Darwin Glass is strongly against a meteorit’ic origin. The evidence for an extra-t’errestrial origin is (I.) the nickel content and the Fe/X ratios which are not typical of terrestrial rocks and (2) the apparent lack of a nearby crater, if Darwin Glass is of impact origin. Further field search is planned. and the areas can be narrowed to those containing argillaceous sandstones.

From t,he da,ta presented in this paper, the follov\-ing conclusions Ilit\-<’been reached. (1) At least two distinct, but closely related variations exist in chemical composition. (2) The K/Na, K/Rb and K/Cs ratios show no evidence of major sclectivc loss during melting. This is interpreted as indicating that. for the cations considered, the present composition is essentially the same as that of t,he source material. (3) Neither Darwin Glass, nor the source material, were produced by terrestrial igneous processes. The Glass is less than one million years old. (4) The major and trace element composition (except for nickel) is consistent with a source resembling an argillaceous sandstone. Such highly differentiated material can originate only on the surface of a terrestrial-type plan&. (5) The Cr/Ni, Ni/Co and Fe/Ni ratios, and the absolute concentration of nickel. are anomalous for terrestrial rocks similar in overall composition to Darwin Glass. (6) The chemical composition is not related to that of the basement rocks. or soils derived from them, at the localities where Darwin Glass is nolv found. This precludes hypotheses involving in situ origins (e.g. as fulgurites). The slag from the Crotty smelters is totally dissimilar in physical appearance. and in chemical composition. (7) The major element composition, and the Fe/Ni ratio, axe similar to that of terrestrial impact glasses. (8) Darwin Glass is not related to the australite group of tektites.

The geochemistry

of Darwin

493

Glass

(9)) The geochemical data presented here are consistent with a terrestrial origin for Darwin Glass by meteorite impact. .~~krio~r,lrrigelrLents-‘l’he authors are grateful to Mrs. A. HALSEY for carrying out the spect.roto Dr. \\-. C'OMPSTON for the strontium isotope data, to Dr. 1. qaplric determinat,ions, ]~~~*I)oI~~:AT.~.for the argon data, and to Dr. K. S. HEIER for critical reading of the manuscript,.

R’EFERENCES _~~~HESS L. H. and FLEISCHER M. (1960) Trace element content of the standard granite (G-l) ilncl st>andard diabase (IV-l). c:.S. Geol. 8urvry Bull. 1113. _~HRE?;S L. H. and TA-FL~R S. K. (1961) Spwtrochemical Analysis. 2nd Edition. Addison\Vesloy, Reading, Mass. 13.4~~~ (:. (1959) Tektites. Mem. -Tat. Mus. l’ict. 23. B.-\sEx. (:. and GASKIN A. .J. (1946) Sabural glass from Macedon, Victoria, and its relationship to ot,her natural gla.ssrs. J. Geol. 54, 88-104. HI.~OS .J. D.. (‘I;LKIN F. and RILEY .J. P. (1959) The abundances of gallium and gwmanilon in terrestrial materials. Geochim. et Cosmochim. Actn 16, 151-180. ('.ZRR iIf. H. nnd TTREKTAS K. K. (1961) Thtx geochemistry of cobalt. Geochim. et C’osrnocl~irn. Acta 23, g-60. (‘ON~)EH H. (1934) Darwin Glass. I&. Xin. ,%nd. du.st. 89, 329--:Uo. 1).%\-1u T. \V. E., SU~~MERS H. S. and AMPT 1:. A. (1927) The Taxnanian

tektite-Tjarwin

(klass.

I’roc. Roy. Sot. l’ict. 39,167-190. I~)BXENH~R~T H. (1957) Untersuchungen xur geochemischen Verteilung des Zirkoniums in tlcr Lithosphiire. Qeochim. et Cosmochjim. Actu 11,279-309. ~HMANN \%.. 11. (1960) Sickel in tektit,es by activation analysis. Geochim. et Cosmochim. Acta 19,149-155. F:HN\NN IV. Il. (1962) The abrmdance of nickel in some natural glasses. Geochim. et Cosmochim. .4 ctn 26, 489-493. F.~IRB.URN H. \jT. it nl. (1951) A cooperative invest,igation of precision, and accuracy in churnical, spcct,rochemical and modal analyses of silicate rocks. U.S. Geol. Swrveg Bull. 980. Fttii~rr~ F. (1960) Beit,rag zur Geochemie des Chroms. Geochim. et Cosmochim. Acta 20, 215-240. HARTMAN M. (1963) Einige geochemische Untersuchungen an Sandst,einen aus Perm tmd Trias. Owc11im. et Cosmochim. Acta 27, 459-499. HICIE:ILk’. S. and TAYLOR S. R. (1959) The distribution of Li, Na. K, Rb, Cs, Pb and Tl in Southern Sorwogian pre-C’ambrian alkali feldspars. Geochim. et Cosmochim. Acta 15, 284.-304. HILLY L. (1915) Darwin Glass, a. new variety of tektit,es. Geol. Survey Tasmania, Rec. 3, l-16. HORSTJ~AN E. L. (1957) The distribution of lithium, rubidium and eaesinm in igneolw and sedimmt,ary rocks. Oeochim. et Cosmochim. Acta 12, 1-28. ISHIH~SHI M. (1959) Cobalt content of shallow wat)er deposits. Kyoto Univ. Inst. Chem. Rc.r. 137rZZ.3’9, 26-38. >IACPHERSON H. G. (1958) A chemical and pet,rographic study of Pre-Cambrian sediments. (kwhirn. et Cosmochism. Acta 14,73-92. MasoN B. H. (1962) Meteorites. Wiley, New York. ~\IIDDLETON G. V. (1960) Chemical composition of sandstones. BUZZ. Geol. Sot. Amer. 71, 1011 1026. PETTIJOHN F. J. (1957) Aedimentary Rocks. 2nd Edition. Harper. PETTIJOHN F. J. (1963) Chemical composition of sandstones. ChaptIer S, Dat,a of Geochemistrv (6th Edition, M. Fleischer editor). U.R. Geol. &.wvey Prof. Paper 440-S. PREUSS E. (1935) Spektralanalytische Untersuchung der Tektite. Chem. d. Erde 9, 365-416. Ross C. S., FOSTER M. D. and MYERS A. T. (1954) Origin of dunites and of olivine-rich inclusions in basaltic rocks. Amer. Min. 39, 693-737. S.&H.&MATH. G. (1945) Spurenelemente der Gesteine im Siidlichen Finnisch-Lappland. Bull. Comm. Ge’oZ. FinZnnde 135, l-86.

494

8. It. TAYLOR and ;\1. SOLoMOX

$2. &I., MIESCH A. T., SEWMAS W. L. and RILEY L. B. (1959) Elemental comPart 3. Geochemistry and mineralogy of the position of the sandstone-type deposits. C’olorado Plateau uranium ores. U.S. Geol. Survey l’rof. Paper 320. &SMITH \T-.C'AMPBELL and HEY M. H. (1952) The silica glass from the crater of Aouello~~l (Adrar, \%‘estern Sahara). Bull. Inst. Emnc. ilfripe Soire, XIV ,I-o. 3, 762-776. S&we, I;ond. 131, 117-118, 876. SPEWER L. J. (1933) Origin of tektites. SPESPEH L. J. (1939) Tektites and silica-glass. *lIiner. Muag. 25, 425-440. Srrtoc~ L. \1-. (1936) Zur Geochemie ties Lithiruns. Sflch~. Ges. Il.iss., (Jhttingen, IV, S.F. 1, x0. 15, 171-204. (&oZ. (ies. Wierl. JIitt. ‘g, SI;ESS F. E. (1914) Hiickschar1 nnd Sencrcs 1il)er die ‘l’ektit,frage. SHOEMAKEE

51-121. SUESS F. K. (1935) Australites. Geol. Xug. Lo&. 72, 288. Gpochinr . et C’osmock km. Acta TAYLOR S. R. (1962) The chemical composition of nnstralites. 26, 68&i_,:! TAYLOR S. H. and SACHS 31. (:eochemical evidence for the origin of arwtralites. Geoch&~. et Cosmochim. dcta 28, 235-264. TAYLOH i3.W. and SOLOMON M. (1962) Geochemical and geological evidence for t,he origin of Darwin Glass. Suture, Land. 196,124-126. TITREKIAX K. K. and \VEDEPOHL K. H. (1961) Dist’riblltion of the elements in some major lmits of the E:art)h’s crust. Bull. Geol. Sot. Amer. 72, 175-192. UREY H. (‘. and GRAIL: H. (1953) The composit)ion of the stone meteorites and the origin of the met,eorites. Geochim. et Cosmochim. Actcc 4, 36-82. VINOGRADOT A. P. and RONOT A. H. (1956) Composition of the sedimentary rocks of- the GeoLhimiya 533-559. Rwsian Platform in relation to the history of its t,ectonic movements. WASHIN(:TOX H. S. (1917) Chemical analyses of igneous rocks. U.S. Geol. Survey. Prof. Puper 99. \VEREI< .J. S. and MIDDLETON C. I--. (196lj Geochemistry of the turbidites of the Normanskill and Charny formations: 1: Effect, of ttlrbidity cnrrent,a on the chemical differentiation of t,urbidites. 1 I: Distribution of trace clcmrnts. Geochiwt. et C’osmochim. Acta 22, 200-243, 244&288. \YEBBER G. R. (1961) Report of Nonmet,allic Standards Committee, (‘anadian Associat,ion for Applied Spectroscopy. Appl. Spect. 15,159-161. lVEI)EPOHL K. H. (1956) Untersuchnngen znr Geochemic des Hleis. (Geochim . et C”osmoch~in~ . Acta 10, 69-148. \~EDEPOHT~ K. H. (1963) 1vritten communication, P.S. 16. c_.is’. Geol. ,%,w~y. T’rof. Paper 440-s.