Chemical Geology. 20(1977) 325--343 325 ~sevier Scientific Publishing Company, Amsterdam -- Printed in The Netherlands

GEOCHEMICAL DISCRIMINATION OF DIFFERENT MAGMA SERIES AND THEIR DIFFERENTIATION PRODUCTS USING IMMOBILE ELEMENTS

J.A. WINCHESTER and P.A. FLOYD Department of Geology, Univermity of Keels, Keele. Staff~ ST5 5BG (Great Britain)

(Received August 18, 1976; revised and accepted October 18, 1976)

ABSTRACT

Winchester, J.A. and Floyd, P.A., 1977. Geochemical discrimination of different malpa8 series and their differentiation products using immobile elements. Chem. Geol., 20: 325--343.

The abundance and distribution of selected minor and trace elements (Ti. Zr, Y, Nb, Ce, Ga and So) in fresh volcanic rocks cam be used to clses/fy the differentiation products of subalkaline and alkaline magma series in a ndmflsr manner to methods using normative or major-element indices. A number of variation diagrams may be used to distinguish common volcanic rock types in terms of the above elementL

As these elements are immobile during po0t-cous~idation alteration and metamorphic processes, this method of rock-type elmmification may, when applied to metsvolcanic rocks, prove more reliable than the commonly used methods that utilize major elements, some of which are known to be mobile.

INTRODUCTION

Chemical criteria commonly used to classify fresh or slightly altered volcanic rocks all use dements known to be mobile during metamorphism. Therefore, they cannot be reliably utilized to identify and classify volcanic rocks which have been metamorphosed or extensively altered. The aim of this paper is to establish a chemical means of discriminating different volcanic rock types and magma series which may be applied to metamorphosed or extensively altered rocks.

This approach necessarily involves much simplification, and is only in- tended as a broad outline of the distribu/~on of immobile trace elements in different volcanic rock types. It is not intended to discuss the relative origins of the rocks and their relationships, nor the reasons for the variations in the trace-element contents. This study attempts to produce a simple graphical representation of immobile element concentrations in volcanic rocks which can be used to recognize the original igneous rock type where metamorphism or alteration has obscured it.

326

The method proposed utilizes the seven minor and trace elements, Ti, Zr, Y, Nb, Ce, Ga and Sc, which are generally considered to remain inert during secondary alteration processes, including spflitization, submarine alteration or metamorphism (Frey et ai., 1968; Cann, 1970; Kay et ai., 1970; Elliott, 1973; Pearce and Cann, 1973; Field and Eiliott, 1974; Herrmann et al., 1974; Pearce, 1975; Ferrara et al., 1976). These elements are referred to below ss "immobile" elements.

In this paper it is intended to set up, using f~esh volcanics a geochemical grid which may subsequently be used as a basis for plotting altered and metamorphosed rocks. In a subsequent paper (Floyd and Winchester, 1978) we intend to illustrate the usefulness of this approach applied to altered and metamorphosed volcanics, ~owing that the original volcanic rock type and the magma series it belonged to may still be recognized by this method in spite of subsequent extensive alteration.

The concentrations of the seven immobile elements and their ratios vary systematically with differentiation in magma series (see Fig.l), and as a result different rock types and magma series may be distinguished by their concentrations or ratios of these elements. In this paper the immobile trace elements are used to discriminate firstly between the main magma series, and secondly between their varied differentiation products. No attempt has been made to distinguish different tectonic regimes of magma emplace- ment.

CLARIFICATION OF ROCK SERIES

The classification presented in terms of immobile elements must clearly correspond to well-known and documented schemes that are in general use.

The main rock series are divided into: (a) alkaline; and (b) subalkaline groups (Chayes, 1965; Wflkiuson, 1968) with the former comprising the alkali olivine-basalt series (Tilley, 1950) and shoshonite association (Joplin, 1965, 1968), and the latter the tholeiitic series (Kennedy, 1988) and the calc-alkah association. Following Wilkinson (1968) it is considered that the high-alumina b,~Its (Kuno, 1960) are representative of the basic members of the calc- alkali association.

On the dis~zms (see Figs. 1--10) broad distinctions are made between degrees of alkalinity (alkaline, subaikaline) and differentiation products (basic, intermediate, acid). Because of insufficient immobile element data, no distinct/on is made between the products of the tholefitic suite and the calc-alkali association. Hence, the suballmline group illustrated here contains both tholefitic and high-alumina basalts, although the andealte and rhyolite fields were delimited using rocks restricted to the calc-aikali association only. Similarly, the rocks belonging to the shoshonitic association have not been plotted on the diagrams, and thus the alkaline group refers to the differentia- tion products of alkali olivine-basalt only. However, the few analyses ob- teined of shoshonitic rocks indicated that the characteristic Zr/TiO2 ratio at

327

least was indistinguishable from that of the alkali olivine-basalt suite, in view of the relatively sodic or potassic nature of the alkali olivine-basalt series (Tilley and Muir, 1964) and the various names applied to each lineage, an attempt was made to distinguish between them using immobile elements. For example, although the volcanic suites of the Tristan Islands (Baker et 81., 1964) and Gough Island (Le Maitre, 1962) are generally considered to be representative of the potassic lineage, with high K20/Na20 ratios throughout, no clear immobile element differences were noted between them and the more sodic lineages.

ROCK NOML-'NCLATURE

Immobile element data shows some overlap between different rock types in most of the accompanying diagrams (see Figs.2, 4, 6, 8--10). This may be expected partly because the natural rock series form a chemical continuum, and partly because the application of a geochemical criterion as a means of distinguishing rock groups classified according to their mineral content must allow for some slight inconsistencies. In addition, there is some variation in the rock nomenclature used by different workers, and for this reason, groups of related igneous rock types have, for simplicity, been collected together under single headings (Table I).

TABLE I

Grouping of rock-~pas used in discrimination diagr.,m and their respective symbols

Subalkaline Mildly alkaline Strongly alkaline

X rhyolite comendite trachyle

pantellerite phonolitic trachyte alkali rhyolite nepheline traehyte

rhyodacite daeite

-

andasite trachyandesite phonolite hifh-K andasite latite trachytic phonolite low-K andes/re benmareite low-Si02 andeslte basaltic ande~te

tholeiite

high-Al basalt

olivine-basalt basanite alkali olivine-basalt trachybaasnite traehybasalt nephelinite hawaiite nepheline---hawaiite mugear i te nepheline--mugearite

328

u ;

- ~ ~ ~

[ _

.i~

| .~ ~. ~= ~,, ,~- ~ ~-~ " ,~c~ ~ , "

m

. . , ,0

8

~O ~O ~Q ~O ~O O0 ~O

U 0

' i i i ' ] j i J r,c.

329

U

o ~ ~ o - o

~ _ ~ o~ - ~~ g~ ~ "~

1 ~ - ~ o o

" " . ! .9

~, .gs

., 1.0

330

Rocks described as latite in the literature are grouped with trachyandesites as no analyses of the intermediate membe~ of the shoshonite association (also termed latlte) are included. Rocks with extreme compositions, such as carbonatites, or members of the lamprophyre suite have not been included in the scheme.

The distinc~ve nature of the variously grouped rock types in terms of absolute abundances and ratios is shown in Table II. Although this compila- tion only uses the data plotted in the diagrams presented here, and is numeri- cally restricted for some groups, progressive trends and differences can be seen within any one rock series.

SOURCES OF DATA

To illustrate how different volcanic rock wpes may be graphically discrim- inated, using the immobile dements (Ti, Zr, Y, Nb, Ce, Ga and Sc), anal- yses obtained from a wide variety of geographical locations and tectonic settings have been plotted on the immobile dement-pair diag~ms (see Figs.l--10). However, as all the dement pairs include a trace dement, none of the many published collections of major-dement analyses of volcanic rocks coultl be used. Also, many collected analyses which included some trace dements contained determinations of only some of the immobile trace elements, and thus could not be used on some of the diagrams. Therefore. a relatively small proportion of the total number of analyses published include analyses of all the elements required for this study.

The data upon which the conclusions of this work are based, therefore, derive from only a few hundred anelyses (Table HI). Consequently, the field boundaries delimited in the following diagrams must be viewed firstly as marking gradations/changes, and secondly as being subject to limited adjust ment as more data becomes available.

DISCRIMINATION DIAGRAMS

It has been known for a long time that trace-element concentrations change in a regular and predictable manner with progressive differentiation (Wager and Mitchell, 1951; Nockolds and Allen, 1954). The accompanying diagrams (see F ip l - - lO) show the systematic variation of selected immobile-element ratios and abundances for various rock groups. They, therefore, represent indices of progressive msgmafic evolution in a similar manner to standard ntajor~lement indices (e.g., mafic index). In addition, the diagrams also demonstrate that a discrimination may be made between the products of subalim]ine (mainly calc-alkali) and alkaline di~erentiation trends. In a pre- vious paper (Floyd and Winchester, 1975), it was shown that discrimination could be made between fresh alkaline snd subalkaline b~]ts by plotting their Zr/P2Os ratio against either thek TiO2 content or Nb/Y ratio. In fresh volcanic rocks the Zr/TiO2 ratio has been applied as a means of discrimination

331

c-. '~"o

. . _ g eam '=" 7 , . . ,

~ . ~ = :~ , . - . - . , - , ca " - " Dr,,~ , ,~ 0

. k I ~.. ~ . -~ ~ ~ 0

,-- , i - . ,~ "~ . . ea ..~ ~.~ ..I ~

l l I I0 ,.-~

0 . v ~ , . . .~'01~1 ~ "O. , .v L',- ~- - O I~L , ,L~.~. ,L~.~. , = l -o~. I

_ -~ ' .=o o=~o.o~.= . . , . , . . . , , .~_~a~.., , ,~=o~

3a 0~

~ "o

~ 3 - ~' " ' - , - -a [ . - ' i a . - i a . .~ _1 . ~ ~ a"a .~"a ~

O. .a ..a m ~e C ,,,, , - , 0

i -3.

o .E

332

between different rock types in preference to the Zr/P~Os ratio because P 15 generally more mobile than the other elements discussed here, as exemplified by its behaviour in a weathering regime (Vogel, 1975).

Si02--Zr/P~Os

Initially the Zr/TiO2 ratio was plotted against SiO2 content (Figs.l, 2), which can be used as a rough measure of the degree of differentiation. With the progressive differentiation of a basic magma, the Zr/TiO2 ratio increases, and reflects the overall decline of TiO2 content in non-basaltic differentiated rocks. This ratio may thus be used as a relatively sensitive differentiation index in place of SiOz. However, a more marked increase in the Zr/TiOz ratio occurs in the more alkaline magma series, reflecting the strong concentration of Zr in alksline rocks. For example, phonolites from the Dunedin Volcano characterktic.~lly have a conskierably higher Zr/TiO2 ratio than the basaltic rocks with which theT are a~ociated, while by contn~t dacitic rocks from Tonga have Z~/TiO2 ratios only slightly higher than those of the basahic andesites with which they are ~r~mciated (F i~ l ) . Thin the amount of in- crease of the Zr/Ti02 ratio in proportion to Si02 increase may be applied as an index of a/kalinitT of a rock suize.

With increasing SiO2 conr~nl the Zr/TiO2 ratios of alkaline and subalkaline suites diverge, and alkaline volcanic rock types plot apart from the su~ka lmt

t,/,~" , . -'rrachyte - T rachyand~te , Benmar~:e "e Alkali Basalt Thl lehtc. High Albmma BI.~III

401 ", Phonolite I~i~lni.be "r.~cl 'yba~ln, te i . . |

om C~ 3 Zr /~02

Fig.1. SiOs--ZriTiOi diagram showing the relative distribution of the differentiation prod- ucti f~om various volcanic centres.

76.

-',2L ; '

- .

- . . j

52. ,. ,~ /

/ ~ t e -- 3ac~e. ~y~lac,te ! :/I ~ d ~ # Comen~te. Pontelier,te

teo

teo

teo

teo

333

~[ RI',YC_-.":

I

72- ~ ; : " / o .n . o

R.,Y:.~C'E -~" e 7 ~ '~ ,' -' 5,6" 3AC -E " ; ~ ' .~ ~ :31vE%31TE "

"-'-:" / " I . I " PA%---L.Eqn'E

~s~- P' ,'.. .... -~-_--J" :". " i . ,..,. _.~" , , -qAC,"YTE _ , >~ , ..,, ~-...:_=/ :'-..,- - _

ec . . . . . - - -- . . . . . . . . & l i ,~-'IlJL~ . . . . "1;:l: TI: ~-'lJ~ ~- - - I I ! - - - - P , - i . J P~" I- M -a . . - -~%. l : _ _ - - J , / . '~ ' _ '~ . ; .--: _- _-_ =_ (~ ~" ~" ' i ' / - - - : . ,~- - : _- - : : : -

~s6- , : . , iT=" ;~ . : - - - : - - _ : :

$2- sus -~,z :~, , ;'. ~ ;&/ . . " - - .~ . : ,ASA.T. 5. ; , ~ ; , - -

, "~ ~l~.''ki'~" un TRACFaYBASA~,T E I

:F4 J

!

"3 01 . . . . . . Zr / Tn0 2

Fig,9-. 8 iO2- -~r / ' r iO s diai~rmum Ihowing tee de l imi ted f ie l tk fo r common vo lean i roc Inn.

Symbols u in FLg.I.

rock types (Fig.l). By using different symbols to denote the various rock types (used consistently on all the accompanying diagrams), the different rock types were found to plot in distinct fields, and approximate boundaries to these fields can be drawn (Fig.2).

8i02--Nb /Y

The Nb/Y ratio was first noted as an indicator of alkalinity in basalts by Pearce and Cann (1973), and further studies confirmed this (Floyd and Winchester, 19"/5); a highe~ Nb/Y ratio generally reflects the higher Nb con- tent cham~t ic of alkaline suites. The Nb/Y ratio was plotted against SiO~ content (Fip.3, 4) in order to establish whether this ratio also tended to change systematically with differentiation. Initially, three magmatic suites were plotted: calc.alkRH~e volcanics from Mr. Ararat (Lamber~ e: al., 1974), a transitional alkaline suite from Easter Island (P.E. Baker et al., 1974), and an a]kA,~e series from the Dunedin Volcano (Price and Taylor, 1973) (Fig.3). With the possible exception of the Mt. Ararat suite, the Nb/Y ratio was found to increase only slightly with increasing SiO2 content: the ratio, therefore, seems to reflect the alkalinity of a magma series alone. Different rock types were found to plot in distinct fields, around which approximate limits could be drawn (Fig.4). Only between a few rock types, notably trachyte, phonolite and trachyandesite are the limits ill-dei'med. For all ex-

teo

334

?E

72

M

52

,o t

O01

-_.. I l l +

: / -

Y i : . "&

? - & >o O. -

n - -. IJJ

" z [

= i j

I I ;

010 10C Nb/Y

v

Fig.3. SiO=--Nb!Y diagram showing the relative distribution of the differentiation pro( uecs from various volcanic eentres. Symbols as in Fig. 1.

~" " "#e te

72 F , COMEN~TE A PANTELLERfrE

RhYODACI"E - " -- % 88 - ~AcirI~ ~ : 0

,. : ~ ~ TRACI-,YTE , , ,, , , ~ - _

": -: : ~

. . . . . _ " ITRACHYANDESITE~

" i ." " - " ! " " ALKALI ~/~"~. . - = ~- =.=,,_. " . . - l~lw.%ftLT I/ ""

= - - "~, - _ " J . . . _ - ' - - - - r

: . . "E" " - "-. :,,'. S,.;B-,~.KALINE BASALT " r.~*. : " . . e.J". BASANITE

:= . - . p; . , ;'~..,'NEm4EUNtTE / I I I I I I

_= j.,,,.,wl I

.,..50-

C~56 L

521-

I

,.8[

~0 ; " , 01 . . . . 0'tC "~00 " "

Nb/Y Fig.4. 8iO2--Nb/Y d/alpram showing the delimited field for common volcanic rocks. Symbols ms in Fig,1.

335

cept the most siliceous rocks a Nb/Y ratio of 0.67 satisfactorily divides generally subalkaline magma suites from those that are alkaline. On addition of further data this ratio was found to produce a better discrimination than the value of 1.0 formerly used by Floyd and Winchester (1975). Peralkaline rock types, such as phonolites, tend to have .~o/Y ratios exceeding 2.0.

Zr /~O2--Nb /Y

Since both the Nb/Y ratio and the Zr/TiO2 ratio are indices of alkalinity, hut only the Zr/TiO= ratio represents a differentiation index, a plot of Nb/Y against Zr/TiO2 was also found to discriminate between different volcanic magma series and rock types. Thus, the calc-alkaline rocks from Mr..Ararat have a low Nb/Y ratio, with the Zr/TiO2 ratio increasing from andesize to rhyolite, while the alkaline rocks from the Dunedin Volcano have a high Nb/Y ratio and a Zr/TiO2 ratio that increases markedly with differentiation (Fig.5). The mildly alkaline suite from l~A~ter Island plots between the two other trends.

1 00r

O-lO

0_ F--

r,,,i I

o-oi L

I I Ic~

,-1,<

I

m, ~ j , ,A .A I

I

i

i

! - -

./."

o-ol 11 Nb/Y 1'oo lO-t) Fig.5. Zr/TiO,--Nb/Y diagram showing the separation of magma series and their respec- tive differentiation products, u illustrated by three volcanic centres. Symbols as in Fig.]..

teo

teo

336

When the other available analyses were plotted different rock types were found to plot in different par6i of the diqpram (Fig.6). Thus subalkaline basalt are c h ~ by relatively low Zr/TiO2 and Nb/Y ratios, whereas dacites and rhyoUtes have a h/gher Zr/TiO2 ra~o while retaining a low .Nb/Y ratio. Alkaline basic rocks, as other basic rocks, are typified by low Zr~iO= rat/os, but have charac~ristical]y high Nb/Y ratios, while alkaline differen- tia~es have typically high Zr/TiO2 and Nb/Y ratios (Fig.6). Thus, major 8roupings of volcanic rock-types may be distingtdshed solely by their characteristic proportions of selected immobile minor and trace elements.

4 i t ' .

4~p v w

(:3 I - -

Im

N

--. .-_

o

t ...ONENDITE ,Ij'_ ~A N'rELLERITE

t " " _ P" ,0 ;~0" - " rE ~- _.

"~ 0 t 0 ~I "- "0 . l amw =, - -

%,

t

qI-'YOL I"E \ 11 -- "~ \ " i l k " . .

q -YOCAC-E f : TRACl-vTE ~ACI'E ' " : - " ~ .

" , " - " ~ . =" ~1~ NEPHEUNP'E - . . . . . " ; - " , ' - - . . - " J L , , . "

" A . . . . S" - " " / " " ' " " " ~'" - ,~mur. r . . / /= _= = =lB ._ " "

I

I I I SI.B -ALKA,.I~IE BASAL"

I

I

I 3 01 0'10 1 '0 lC O

Nb/Y Fil~& Zr/TiOz--NblY rli=-~ram ihowinll the delimited fields for eOmmnn volcanic rocks. The Zr/TiO= ratio acts as a differentiation index and the NblY ratio as an alkel/nity index Symbo Is as in Fig, 1.

337

Ce--Zr/~02

The contents of additional trace elements have been compared in different volcanic rock types in a similar manner. The Ce content in differentiated rock types of the calc-alkaline association remained broadly constant, where- as in alkaline rocks Ce content was found to increase markedly with differen- tiation. In a disgram plotting Ce content against the Zr/TiO2 ratio, subalkeline and alkaline trends showed a marked divergence (Fig.7).

When all available data was plotted, the rock types were found to plot in distinct fields (Fig.8). Rather more overlap occurred than in previous dia- grams; for example different basaltic types and basanites were not clearly distinguished, and the scarcity of data implies that this plot may not be quite as reliable as former ones.

'L

-t

0 -

ZT.

/ / "!: /

I I 250 30C 350

Ce pp.m

Fig.'/. Zr/TiO~--Ce diallram showing the relat/ve distribution of the differentiat/on products from various volcanic centres. Symbols as in FiK.1.

338

- .?=-

I I !

__ lc -_ _% - -

- : -~ '~OL "E ~- ' - - _ ".= - -= A-" -~-_~

t

o: _.- . . . t -~- - i . . f . : _ . _ . : . :.--go.-.

- ,,,---ip e Zp~. . - S_- ~.A_-

Ce Dpm

Fig.8. Zr/TiOF-Ce diagram showing the delimited fields for common volcanic cocks. Symbols as in Fig.1.

Ga--Zr/Ti02

A similar pattern was found in plots of Ga a~inst ZrlTiO2 ratio (Fig.9). Whereas alkaline rocks show a concentration of Ga with differentiaUon, subalkalJne rocks show little change, or possibly a slight decrease of Ga content. Again the main volcanic rock types group in different fields, al- though there is considerable overlap of basaltic types, between basalt and andesite, and between phonolite and pantellerite.

Ga/Sc--Nb /Y

Sc contents are markedly reduced with differentiation in both subalkaline and alkaline rocks. Thus the .Ga/Sc ratio may be employed as an index of differentiation, so that in rocks belonging to the calc-alkaline association

339

' / /PANTB.LERITE " / I [] ~,. / . ,= , /

I / '

I ' / RHYOLITE / , /

! , I / PHONOLITE [ / TRACHYTE

I : / l l X X 0-10 " " / / a

J " RHYODACITE I +4- : DAClTE " / t ~ ~

(~ A ; + i : " . . . . . / TRACHYANDESITE

i ...~.~...:-,"!-, ! OC.. ......--" . I . " . = I "

- - " : ; i i 1 : = I :

' " =. . - ~: ; ' ' " . = .

" , ALKALINE BASALT

SUB -ALKALINE I

I i i i i i i

5 10 15 20 25 3C 35 0 &5

Go ppm

Fi~9. Zr/TiOz--Ga diagram showing the tentatively delimited fields for common volcanic roelm Symbols as in Fig.l.

Ga/Sc ratios increase with differentiation from andesite to rhyolite; where- as rocks of the alkaline magma series also show an increase in the Ga/Sc ratio, although less marked. PanteUerites, containing very low Sc contents are distinguished by very high Ga/Sc ratios. In Fig.10 the Ga/Sc ratio is plotted against the Nb/Y ratio, used as an index of alkalinity. As before, the main volcanic rock types plot in distinct fields: in particular there is a clear separa- tion between the subalkaUne and alkaline basalts. The data on which this diagram is based is particularly scarce, and it seems likely that, with the

340

d= ~ i

:~t-YOL T=> ..---;-, i I

i I

U IV l J ~"

; " " / ~OhC, . TE : j J ' ,~=. , . I "RACi-'Y"E

:_,:--- T.*C.YAN S,TC =, I I I .

. . . )~ . . : _ .= :

33"ANDL:'S'E : : q F ' ' ~-~- -~ - " " 'A.KAL'

/ "~" == * " =' ",L -- BASALT

- - =. f" -. ~ '~4 ~ P ' " i

SUS-A_KAL I~ / BASA="

J~

I |

~'~ Nb/Y Ioo

Fig.lO. GaI~--Nb/Y diagram Ihowinl the approzimte distribution of common volcanic rockL Note the divergence of the alkaline and sub-alkaline masrna series.

accumulation of further analyses, some of the compositional fields plotted on Fig.lO may be cormidembly enlarged, although the fundamental distribu- tion pattern is unlikely to alter.

CONCLU~ONS

Figs. 1--10 illustrate that contents or ratios of seven trace elements are charactm~tic of certain volcanic rock types, and that their concentrations are strongly controUed d .urJng differentiation. By using fresh volcanic rocks a number of pochemical grids can be devised so that ff a suite of fresh volcanic rocks is plotted the identity o f the different rock types present can be deduced. More ~ i~mt~y, as the elements used are immobile during metamorphism, these pochemlcal grids should also be applicable to-exten- sively altered and metamorphosed volcanics. They should thus provide a

341

more reliable method o f recogniz ing the original volcanic rock type where ext reme alterat ion or metamorph ism hss taken place.

REFERENCES

Appleton, J.D., 1972. Petrogenesis of potaasium-rich lavse from the Roecamonfino Volcano, Roman region, Italy. ,I. Petrol., 13: 425--45~

Bailey, D.K. and MacDonald, R.J., 1970. Petrochemical variations among mildly per- alkaline (eomendite) obsidians from the oceans and continents. Contrib. Mineral Petrol., 28: 340--351.

Baker, L, 1969. PetroloID" of the volcanic rocks of Saint Helena Island, South Atlantic. Bull. Geol. Sop. Am., 80: 1283--1310.

Baker, P.E., 1968. PetxoloiD" of Mt. Misery Volcano, St. IQtts, West Indies. Lithos, 1: 124--150.

Baker, P.E., Gass, LG., Harris, P.G. and Le Maitre, R.W., 1964. The volcanolosieal report of the Royal Society Expedition to Tristan da Cunha, 1962. Phflo Trans. R. 8oc. London, Ser. A, 256: 489--575.

Baker, P.F~, Broaset, R., Gus, LG. and Neary, C.R., 1973. Jebel al Abyad: a recent alkalic volcanic complex in western 8andi Arabia. Lithos, 6: 291--308.

Baker, P.E., Buckley, F. and Holland, J.G., 1974. PetroloID" and geochemistry of Easter Island. Contxib. Mineral. Petrol., 44: 55--100.

Cann, J.R., 1970. Rb, 8r, Y, Zr and Nb in some ocean floor basaltic rocks. Earth Planet. Sd. Lett., 10: 7--11.

Cheyas, F., 1965. Titania and alumina eontent of oceanic and cireumoeesulc basalt. Mineral. Mall., $4: 126--181.

Cox, K.G. and Hornung, G., 1966. The petrology of the Karroo basalts of Basutoland. Am. Mineral., 61: 1414--1432.

k'l-Hinnawi, E.E., Pichler, H. and Zeil, W., 1969. Trace element distribution in Chilean ipimbriteL Contrib. MineraL Petrol., 24: 6C ~,2.

Rlliott, R.B., 1978. The chemistry of gabbro/amphibolite ~analtiorm in south Norway. Contrib. Mineral Petrol., $8: 71--79.

Engei, A.~J., Enpl , C.G. and Havens, R.G., 1965. Chemical characteristics of oceanic besalts and the upper mantle. Bull. Geol. Sop. Am., 76: 719--764.

Ewart, A. and Bryant, W.B., 1972. Petrography and geochemistry of the igneous rocks from Eua, Tongan Islands. Bull. Geol. SoP- Am., 83: 3281--3298.

Ewart, A., Taylor, S.R. and Capp, A.C., 1968a. Trace and minor element geochemistry of the rhyolitic volcanic roclm, central North Island, New Zealand. Contrib. Mineral. PetroL, 18: 76--104.

Ewart, A., Taylor, 8.R. and Capp, A.C., 1968b. The geochemistry of the pantellerites of Mayor Island, New Zealand. Contrib. Mineral. PetroL, 17: 116--140.

Ewert, A., Bryant, W.B. and Gibb, J.B., 1973. Mineraloi~ and geochemistry of the young mr volcanic islands of Tonga, southwest Pacific. J. Petrol., 14: 429--46~

Ferrara, G., Innoeanti, F., Rieci, C.A. and Serri, G., 1976. Ocean floor affinity of bsealts from north Apennine ophiolites: geochemical evidence. Chem. Geol., 17: 101--111.

Field, D. and mliott, ILB., 1974. The chemistry of gabbro/amphibolite txansitions in south Norway. Contrib. MineraL Petrol., 47: 68--76.

Finwer, M.F.J., 1975. 'Itrace element distribution in lavas from Anjouan and Grande Comore, western Indian Ocean. chem. Geol., 12: 81--98.

Floyd, P.A. and Winchester, J.A., 1975. Magma type and tectonic setting discrimination using immobile element& Earth Planet. SoL Lett., 27: 211--218.

Floyd, P.A. and Winchester, J.A., 1978. Identification and discrimination of altered and metamorphosed voleanie rocks using immobile alement Chem. Geol., 21 (in press).

342

Frey, F.A., Haskin, M.A., Poetz, J.A. and l-laskin, L.A., 1968. Rare earth abundances in some basic rocks. J. Geophys. Res., 73: 6085--6098.

Gass, I.G. and Mailiek, D.I.J., 1968. Jebel Khariz: an Upper Miocene strato-voleano ot comenditic affinity on the South Arabian coast. Bull. Voleanol., 32: 33- 88

Gasa, I.G., Mailick, D.LJ. and Cox, K.G., 1973. Volcanic islands of the Red Sea J. Geol. Soe. London, 129: 275--310.

Gibson, LL., 1970. A pantelleritic welded ash-now tuff from the Ethiopian Ril'1 Valle_~. Contrib. Mineral. Petrol., 28: 89--111.

Gibson, LL., 1972. The chemistry and petrogenesk of a suite of pantelleritas rom the Ethiopian Rift. ,I. Petrol., 13: 31--44.

Gill, J.B., 1970. Geochemistry of Vi$i Levu, Fiji, and its e~)lution as an island arc. Con- trib. Mineral. Petrol., 27: 179-203.

Goldich, S.S., Treves, S.B~, Sahr, N.I-L and Stucklass, J.S., 1975. Geochemistry of the Cenozoic volcanic rocks o[ P~oss Island and vicinity, .~mtaretiea. J. Geol., 83 415-435.

Heming, I~F., 1974. Geology and peizoloBy of Rabaul Caldera, Papua New Guinea. Bull. Geol. SOe. Ar~, 65: 1253--1264.

Herrmann, A.G., Ports, M.J. and Knake, D., 1974. Geochemistry of the rare earth ele- ments in spilites from the oceanic and continental crust. Contrib. Mineral. Petrol.. 44: 1--16.

Joplin, G.A., 1965. The problem of the polish-rich basaltic rocks. Mineral. Mag., 34: 266--275.

Joplin, G.A.: 1968. The shushonite mmociation: a review. ,I. Geol. SOc. Aust., 15: 275-- 294.

Kay, R., Hubbard, N.J. and Gast, P.W., 1970. Chemical characteristics and origin of oceanic ridae volcanic rocks. O. GeophyL ReL, 75: 1585--1614.

Kennedy, W.Q., 1933. Trends of differentiation in basaitic re,areas. Am. J. Sci., 215: 239--256.

Keason, S.E., 1973. The primary geochemistry of the Monaro alkaline volcanics, south- east Australia -- Evidence for upper mantle heterogeneity. Contrib. Mineral. Petrol., 42: 93--108.

Kuno, H., 1960. lligh-alumina basalt. J. Petrol., 1: 121--145. Lambert, R., St.J., Holland, J.G. and Owen, P.F., 1974. Chemical petrology of a suite of

caie-aikaiine laves from Mount Ararat, Turkey. J. Geol., 82: 419--438. Le Maitre, R.W., 1962. Petrology of volcanic rocks, Cough Island, South Atlantic. Bull.

GeoL Soc. Am., 78: 1309-1340. Lipman, P.W. and Moench, R.H., 1972. Bamdts of the Mount Taylor volcanic fieM, New

Mexico. Bull. Geol. SO,- Am., 83: 1335--1344. Lippard, 8.J., 1973. Plateau phonolite lava flows, Kenya. Geol. Meg., I I0 : 543-549. Lowder, G.G., 1973. Late Cenozoic transitional alkali olivine-tholeiite basalt and andesit~

from the margin of the Great Basin, southwest Utah. Bull. Geol. Soe. Am., 84- 2993-3012.

Lowder, G.G. and Carmichael, I.S-E., 1970. The volcanoes and caldera of Taissea, Kew Britain: Geology and petrology. Bull. Geol. Soe. Am., 81: 17--33.

Nash, W.P., Carmichael, LS.E. and Johnson, R.W., 1969. The mineralogy and petrology of Mount Suswa, Kenya. J. Petrol., 10: 409-439.

Noble, D.C. and Parker, D.F., 1974. Peraikaline silicie voleanie rocks of the western United StateL Bull. Voleanol., 83: 605--828.

Noble, D.C., Korrinp, M.K., Hedge, C.E. and Riddle, G.O., 1972. Highly differentiated subalkaline rhyolite from Glass Mountain, Mono County, California. Bull. Geol. SOe. Am., 83: 1179--1184.

Nockolds, S.R. and Allen, R., 1954. Geochemistry of some igneous rock series, II. Geo- chim. Cosmochin~- Acta, 5: 245--285.

Pearce, J.A., 1975. Basalt geochemistry used to investigate post tectonic environments in Cypru& Tectonophysics, 25" 41--68.

343

Pearce, J.A., 1976. Statistical analysis of major element patterns in basaltL J. Petrol., 17: 15--4X

Pearce, J.A. and Cann, J.R., 1978. Tectonic setting of basic volcanic rocks determined using trace element analyseL Earth Planet. Sel. Lett., 19: 290--300.

Price, R.C. and Taylor, S.R., 19"/3. The geochemistry of the Dunedin Volcano, East OtaLo, New Zealand: raze earth E]ementL Contrib. Mineral. Petrol., 40: 195--206.

Ridley, W.L, 19"/0. The petroloiD" of Las Canadas Volcanoes, Tenerife, Canary Islands. Contrib. Mineral. Petrol., 26: 124--160.

Strong, D.F., 1972a. The petrology of the lavas of Grande Comore. J. Petrol., 13: 181--217.

Strong, D.F., 1972b. The petrology of Moheli, western Indian Ocean. Bull. Geol. Soc. Am., 83: 389--406.

Taylor, S.R. and White, A.J.R., 1966. Trace element abundances in andesites. BulL Volcanol., 29: 177--198.

Taylor, S.R., Ewart, A. and Capp, A.C., 1968. Leucogranita and rhyoUtes: trace element evidence fuz fractional crystallization and partial melting. Lithos, 1: 1"/0-186.

Taylor, S.R., C.app, A.C., Graham, A.L. and Blake, D.H., 1969. Trace element abundances in andesites, II: 8alpan, Bougainvflle and Fiji. Contrib. Mineral. Petrol., 23: 1--26.

Tilley, C.E., 1950. Some aspects of mngmatic evolution. Q.J. GeoL Soc. London, 106" 37--61.

Tilley, C.E. and Muir, LD., 1964. Oceanic basalt-~aehyte association. Geol. F6ren. Stockholm F~rh., 85: 436--444.

Vogel, D.E., 1975. Precambrlan weathering in acid metavolcan/c roc~,, from the Superior Province, V'dlebon Township, south~entral Quebec. Can. J. Earth 8ci., 121 2080-2085.

Wager, L.R., 1951. The distribution of trece elements during strong i~rectionation of basic magma: a further study of the Skaerijaard intrusion, E. Greenland. Geochim. Cosmochim. Acta, I: 129-208.

Weaver, S.D., 8ceal, J.S.C. and Gibson, LL., 1972. Trace element data relevant to the origin of trachytic and pantelleritic lavas in the East African Rift system. Contrib. Mineral. Petrol., 86: 181--194.

W~Ikinson, J.F.G., 1968. The petrography of basaltic rocks. In: ILH. Hess and A. Polder- vaart (Editors~ Basalts: The Poldervaart Treatise on Rocks of Basaltic Compositions, Vol.l, W'dey, New York, N.Y., pp. 163--214.

Zeil, W. and Pichler, I-L, 1967. Die k//nozoische Rhyolith-Formation im mitderen Ab- schnitt der Anden. GeoL Rundseh., 57: 48--81.