American Mineralogkt, Volume 64, pages 788-798, 1979

The structure of vanadium-bearing tourmaline and its implications regarding tourmalinesolid solutions

FRaNrrtN F. FoIT, Jn. eNo Putltp E. RosnNseRc

Department of Geology, ll/ashington State UniversityPullman, llashington 99 I 64

Abstract

The structure of an alkali cation-deflcient, vanadium tourmaline has been refined (R :0.Ml for 2727 rntensity data) in order to evaluate the effects of tourmaline composition onstructural distortion. V-tourmaline, (Nao*Ca.r"Mgo.,rIGor)(Mg,rrVlj"Ct'olrFe2o:r8) (Al5s6V3*38Tit*o6XSi563Abj7XB2e8&or)Orr(O, roOH258Fo32), is rhombohedral with a : 15.967(2\and c: 7.191(l)A; space group R3m. Despite compositional di-fferences the structure is verysimilar to those of other members of the tourmaline group, most notably a recently refinedaluminous dravite. Analysis of tourmaline-group structural data reveals (l) a negative corre-lation between tetrahedral bond angle variance and mean "alkali'(3a)-oxygen bond length;(2) negative correlations between both 9b and l8c octahedral angle variances and mean 9b-Obond length; (3) a negative correlation between ditrigonality and 9b octahedral angle vari-ance; (4) a positive correlation between weighted mean octahedral bond length and cell vol-ume; and (5) a positive correlation between Na occupancy in the 3a site and mean 3a-O bondlength. These observations demonstrate a systematic flexibility of the structure in response todiverse cation substitution.

The lack of a distinct coupling between the sizes of the 9b and l8c octahedra in refinedtourmaline structures and the extensive and possibly complete substitution of Al3* in the 9bsite in tourmalines suggest that the presence of cations that can vary in size (e.g.,

""2+' r+

) inorder to create a compatible edge may not be a prerequisite for dravite-elbaite solid solution.In view of the structural flexibility of tourmaline and the ease of proton exchange to maintaincharge balance, the apparent immiscibility of dravite and elbaite is thought to reflect extremefractionation of Mg and Li during petrogenesis and by the tourmaline structure, due to thelarge difference in the field strengths of these cations. Other major features of tourmaline sub-stitutional chemistry are also rationalized on the basis of cation field strength.

IntroductionThe crystal chemistry of the tourmaline group is

exceedingly complex. In addition to solid-solution se-ries extending from schorl to dravite and from schorlto elbaite, Foit and Rosenberg (1974,1975,1977)re-ported the existence of two additional substitutions(l) R* + R2* : R'* + E and (2) H* + R2* : R3* innatural tourmalines of the schorl-dravite series,which together result in solid solution toward an al-kali cation-defi cient series, Ri_.Rl*R* ( B O, ), S i. O,,o._,(oH),*..

The extent of substitution in any structure is deter-mined not only by the coordination and dimensionsof the sites available for substitution and the toler-ance of the structure to distortion, but also by thecharacteristics of the ions themselves (e.g., field

w3-6+X/79/0708-0788$02.00 788

strength). While alarge amount of structural data onmembers of the tourmaline group (dravite, Buergeret al.,1962; buergerite, Barton, 1969; elbaite, Donnayand Barton, 1972; schorl, Fortier and Donnay,19751,uvite and an aluminous dravite, Schmetzer, 1978)has accumulated over the past 15 years, systematicstructural changes produced by these substitutionshave not been deciphered. The structure of a signifi-cantly alkali cation-deficient, vanadium-rich tourma-line has been examhed and analyzed in conjunctionwith all available tourmaline structural data in an at-tempt to evaluate the structural signifisxlce of cationsubstitution. It was anticipated that this study wouldlead to a better understanding of the role of struc-tural distortion and cation field strength (Weyl andMarboe, 1962) n limiting tounnaline composition.

FOIT AND ROSENBERG:

Vanadium-bearing tourmaline

Specimen desciption and chernistry

Snetsinger (1966) reported the occurrence of anunusually vanadium-rich tourmaline so€xisling withqvartz, graphite, barium-vanadium muscovite, andbiotite in a quartz graphite schist at Silver Knob,Mariposa County, California. The crystals werefound to be compositionally zoned with cores appre-ciably deficient in alkali (alkaline earth) cations.

A large number of tourmaline fragments were sep-arated, using heavy liquids, from a ground handspecirnen of schist from this locality. Although crys-tals suitable for X-ray analysis were difficult to iso-late due to the abundance and fine dispersal ofgraphite throughout the rock, an equant (-0.3 mm indiameter), optically uniform fragment of the corezone was finally extracted from the tourmaline frac-tion.

An electron microprobe analysis of this crystal isgiven in Table l. Analyses made at many points onthe crystal showed it to be essentially homogeneous.Only one element, V, showed a statistically signifi'cant variation, and this corresponded to a decrease ofless than 0.25 weight percent at one edge of the crys-tal. Vanadium and the small amount of iron presentwere initially assumed to be in the 3* and 2+ va-lence states, respectively. Water was obtained by dif-ference. Since increasing the valence states of ironand/or vanadium reduces the apparent amount ofwater, the amount indicated in Table I represents a641imurl value for this crystal. A comparison of cal-culated and ideal anionic valence sums for the van-adium tourmaline structure (see Discussion) suggeststhat the actual proton content may be slightly lower.

Despite the deficiency of alkali cations, the alkalisite (3a) appears to be fully occupied. An excess ofcations (Mgt*, Fe'*, Cr3*, Vt*, Tio*) normally foundin octahedral coordination which is equal to the al-kali cation deficiency strongly suggests that Mg'z* (or,less likely, Fe'z*) partially occupies the alkali site(Table l).

Unit cell and intensity data collection

X-ray photographs displayed diffraction symmetryand lattice extinctions compatible with space groupR3zr. All reflections were sharp; no weak or diffuse"extra" reflections were observed on long-exposurephotographs. The unit-cell parametem, 4 : 15.967(2) and c: 7.191(l)A, were obtained by least-squaresrefinement of 12 high-angle reflections collected us-

YAN A DI U M.BEA RI NG TO URMALI N E

Table 1. Vanadium tourmaline. Microprobe analysis

wr. percent* Numbers of ions

789

oxid€

s io2 3 3 . 6 1

B 2 o 3 1 0 . 3 2 . 9 4

A 1 2 0 3 3 0 . 0 3 5 . 9 3

C r 2 o 3 1 . 4 8 0 . 1 9

v z o 3 a , 5 2 1 . 1 4

M g O A . 2 3 2 . O 5

T i o 2 o . 4 5 0 . 0 6

F e o I . 2 5 0 . 1 8

M n O 0 . 0 0 0 ' 0 0

N a 2 o I . 3 4 o . 4 4

C a O 2 . 0 2 0 . 3 6

o . o z

Q c a r c = 3 . 1 6 s9 7 . 6 9

H z O 2 . 3 1

1 0 0 . 0 0

F o r m u l a o n t h e b a s i s o f 3 1 ( o , o H , F ) ; ( N " 0 . 4 4 C " 0 . 3 6 M 8 0 . 1 6 K 6 . 6 2 ) ( M e 1 . g 7

vfrruc. l ] r r r " f r+r , ) terr . ruv! ] rur i f f06) (s is.63A10.37)B2. e8Eo. o20z7or. lo

(oHz . saFo . : z )

* A r

" v e . r g e o f a n a l y s e s a l 5 d i f f e r e n c l o c a t i o n s o n t h e c r y s t a l -

ing a computer controlled Picker 4-ctrcle difrac-tometer and MoKa radiation.

A total of 2727 reflection intensities was collectedusing a 0.05" step scan size, a 2-second step count,and a 20-second background count. While 1645 te-flections were collected in the 60" a! al c* wedge overa range 20:5-90", only 1082 were collected in the-at-a!-c* wedge, due to a diffractometer arrange-ment limiting the accessible 20 rcgion to 5-75".Three standard reflections representing a wide rangeof intensity were measured after every 50 reflectionsin order to monitor long-term diffractometer per-formance. The data were corrected for Lorentz andpolarization effects, but absorption effects were ne-glected because of the approximate spherical shapeof the crystal and its low linear absorption coefficient(p : 16.8 cm-'). Subsequent refinement was carriedout using the 2602 structure factors for which F"o. >3o(F""J.

Refinement

Least-squares refinement was undertaken using amodified versions of Onrrs written by Busing et a/.

5 . 6 3

K " O O . I I

9 7 . 9 4

L e s s O = F - 0 . 2 5

790 FOIT AND ROSENBERG: VANADIUM.BEARING T0URMALINE

Table 3' vanadium tourmaline. Atom positional and thermal parameters (standard deviations in parentheses)

A tom P o s i t i o n v

0

0 . 1 2 3 s 3 ( 3 )

o.29782(3)

0 . 1 0 9 7 6 ( 7 )

0 . 1 9 1 7 3 ( 2 )

0

0 .06085 (5 )

0 . 2 6 3 8 7 ( 1 3 )

0 . 09306 (6 )

o 18444 (13 )

0 .19548(7 )

0 . 2 8 5 5 9 ( 7 )

o . 2 o 9 o 3 ( 7 )

0 . 26 rs ( 21 )

0

x / 2

o.26152 (3)

2x

0 . 18 990 (3 )

0

2x

x l 2

2x

x / 2

0 . 1 8 s 7 6 ( 7 )

0 . 2 8 s 3 6 ( 7 )

o.26984(7)

x l 2

o .2r7 64 (2L)

o . 63364 (8 )

0 . 6 1 0 1 1 ( 8 )

0 .45 1 78 (30)

0

0 . 77 026 (39 )

o.48r4o(24)

0 . s 0 9 1 9 ( 2 3 )

o .07162 (23 )

0.092s6(22)

0 .77616 ( r 5 )

0 . 0 7 8 0 3 ( 1 5 )

o .4387 9 ( r 7 )

0 . 3901 (5 7 )

1 . 5 0 ( 3 )

0.446(9)

0 . 3 s 3 ( 7 )

0 . 4 0 ( 3 )

0 . 3 3 3 ( 6 )

o . 7 2 ( 4 )

0 . 7 9 ( 3 )

0 . 7 8 ( 2 )

0 . 7 6 ( 3 )

0 . 7 8 ( 3 )

o .54 (2 '

0 . 6 1 ( 2 )

0 . 60 (2 )

- 0 . 1 7 ( s 7 )

196 (3 )

58 (2 )

44(7)

46 (5 )

3 e ( 1 )

113 (6 )

ee<4 )

r87 (7)

7s(4)

180 (7 )

87 (4 )

73(4)

36 (3 )

Brr

4 2 ( t )

s1 (1 )

4 r ( 7 ,

3 e ( 1 )

R-11

48 (5 )

104 (4 )

r57 (7)

7 1 ( 3 )

82(4)

63 (3 )

80 (4 )

670 (20)

306 (6)

181 ( s )

277 (26)

192 (5 )

341 (38)

s4L(24)

242(r9)

398 (22)

42O(22)

2r8 (r2)

2e4(r4)

484 (1 5)

0

-47 (3)

| ( 2 )

Bn l2

- 2 ( 2 )

0

Bn l2

11 ( r 0 )

Bn /2

4 (10 )

-8 ( 6 )

-30 (6 )

19 (6 )

BD t2

11 (2 )

- 2 ( 2 )

-e(2)

0

22 (9 )

3 . ^ / 2

-28 ( r 0 )

873/ 2

- 2 L ( 6 )

- 15 (6 )

40 (5 )

B B R'72 R- 13Na , Ca ,Mg

M g , V , C r , F e

A 1 , V , T i

B

s i , A I

0 ( 1 )

o (2 )

o (3 )

o (4)

o ( 5 )

o (6 )

o ( 7 )

o ( 8 )

H b ( 3 t

3a

9b

1 8 c

9b

1 8 c

9b

9b

9b

9 b

1 8 c

1 8 c

I8c

9b

Bt t l 2

B:_:./2

2 2 ( r )

Bzz l2

i 8 ( i )

Bn l2

822/ 2

Btt/'

822/ 2

ltt/ 2

44(3)

14 (3 )

27 (3 )

r s o t r o p i c t e m P e r a t u r e f a c t o r s a r e f r o m t h e l a s t c y c l e o f r e f l n e n e n t b e f o r e c h a n g i n g t o a n i s o t r o p i c t e m p e r a t u r e f a c t o r a .

x I O J

(1962). Initial position parameters were those ofbuergerite (Barton, 1969). The form factor tableswere fonnulated from the data of Cromer and Waber(1965) for neutral atoms and the anomalous dis-persion coemcients given in International Tables forX-ray Crystallography. The observed structure fac-tors (,() were weighted on the basis of l/dowhere o.is the standard deviation of ,( based on counting sta-tistics.

The weighted residual (R*) converged to a value of0.060 after several cycles of refinement during whichthe scale factor, atom coordinates, and isotropic tem-perature factors were varied. Comparison of ratioF"(sll)/F"(sll) to that observed for buergerite, fromwhich the absolute orientation of the structure hadbeen determined (Barton, 1969), indicated that theindexing of the vanadium tourmaline data set neededto be transformed (hkl ---> kh-I) to be compatible withthe conventional orientation. Subsequent refinementu5ing the transformed data set reduced R* to 0.042and resulted in reasonable isotropic temperature fac-tors for all atoms.

The cation occupancy ofthc l8c and 9b octahedralsites was determined by monitoring the residual andtemperature factors as the relative proportions of"light" (L : 0.60A1 + 0.40Mg) and "heavy" (Hv:

0.69V + 0.l6Cr + 0.l3Fe + 0.02Ti) atoms were var-ied. Minima in the residuals (R : 0.041 srld R* :O.0/.7) and isotropic temperature factors of 0.35 and0.45 for the l8c and 9b "atoms," respectively (Table3), were achieved with distributions corresponding tol8c : Lo"rHvoo, and 9b : Lo."rHvo.rr.

Several cycles ofleast squares, during which posi-tion and anisotropic thermal parameters and scalefactor were varied, further reduced the residuals Rand R* to 0.029 and 0.035, respectively. Refinementof the proton positions was then undertaken, usingthe positional parameters determined from neutrondiffraction studies of buergerite (Tippe and Hamil-ton, l97l). However, only the positional and iso-tropic thermal parameters of the hydrogen atombonded to O(3) could be varied successfully. This isdue, no doubt, to its higher multiplicity compared tothe hydrogen atom presumed to be bonded to theO(l) oxygen atom. The refinement was completed bysimultaneously varying all positional and anisotropicthermal parameters (except for hydrogen, which wasvaried isotropically). The final residuals are R :0.028, R* : 0.034 for the 2602 data used in the re-finement and R: 0.033, R* : 0.041 for the 2727 datacollected (Table 2'). The positional and thermal pa-rameters are fisted in Table 3. The dimensions and

FOIT AND ROSENBERG: VANADIUM-BEARING TOURMALINE 791

orientations of the thermal ellipsoids (Table 4') andthe interatomic distances and angles (Table 5, uncor-rected for thermal motion) were calculalsd u5ing theFortran function and error program ORrrn (Businget al.,1964).

Discussion

Vanadium tourmaline structure and its relationship toother refined tourmaline structures

The structure of vanadium tourmaline is very simi-lar to those of other refined tourmaline structures(dravite, Buerger et al., 1962; buergerite, Barton,1969; elbaite, Donnay and Barton, 1972; schorl, For-tier and Donnay, 1975; uvite and aluminous dravite,Schmetzer, 1978), most notably aluminous dravite.For this reason, only the details of the structure andstructural comparisons will be given. The reader isreferred to two excellent diagrams (Buerger et al.,1962; Barton, 1969) which portray the general config-uration of the structure.

Tetrahedral ring. The mean T-O bond length(1.625A) in vanadium tourmaline is the largest yetobserved for a tourmaline structure. This is to be ex-pected, since Al'*-for-Sia* substitution (Sir.urAlo.r) isgreater in vanadium tourmaline than in any othernatural lsumaline structure refined to date. Theonly other tourmaline with a significant amount ofAl'*-for-Sio* substitution, dravite (Buerger et al.,1962), has the second largest mean T-O bond length( l .62lA).

Variances in the O-T-O angles druo,, which havebeen used as a measure of distortion (Robinson e/ c/.,l97l), arc apparently related to the mean "alkali cat-ion" (3a-site cation)-oxygen bond length; they grad-ually decrease with increasing mean "alkali"-oxygenbond length (Fig. l). This is consistent with the factthat each 3a coordination polyhedron shares the sixcontiguous interior basal edges of the ring. Thus, anyvariation in the size of the cation in the 3a site shouldaffect the configuration of the ring andlor its individ-ual tetrahedra. A similar coupling of tetrahedral dis-tortion to effective [E cation su;e (Ca'*, Mg'*, Fe'*,Mn'*) was observed in the garnets (Robinson et a/.,l97l). While use of angular variance for dravitemight be questioned on the grounds that dravite mayhave been refined in the wrong absolute orientation

t To obtain a copy of Tables 2 and 4, order Document AM-79-106 from the Business Office, Mineralogical Society of America,2000 Florida Ave., NW, Washingtorq DC 20009. Please remit$1.00 in advance for the microfiche.

Table 5. Vanadium tourmaline. Interatomic distances and angles(standard deviations in parentheses)

Dis tance (A) A loos Dls tance (A) . lng les ( " )

s i - o ( 4 ) 1 . 6 3 0 ( 1 )s i l o ( s ) 1 . 6 4 6 ( 1 )s i - 0 ( 6 ) 1 . 6 1 3 ( l )s i - 0 ( 7 ) 1 . 6 1 2 ( 1 )

Mean 7.625

M e a n 2 . 6 5 3

o(4) - o (5)o(4) - 0 (7)0 ( 5 ) - 0 ( 7 )o(o t - o t+)1o ( 6 ) - o ( s ) *0 ( 6 ) - o ( i )

2 . 5 6 7 ( 2 ) 1 0 3 . 1 9 ( 7 )2 , 6 6 3 ( t ) 1 r 0 . 3 1 ( 7 )2 , 6 7 4 ( t ) 1 1 0 . 4 5 ( 6 )2 . 6 8 3 ( 2 ) r 1 1 . 6 4 ( 7 )2 . 6 7 9 ( 2 ) 1 1 0 . 5 6 ( 7 )2 . 6 5 0 ( 1 ) 1 1 0 . 4 9 ( s )

B-0(2)s-o(s) l

Mean

Ms-0(1)vs-oQ)2Ms-0 (3) ̂Itg-0 ( 6) r

Mean

A1-0 (3)A1-0 (6 )er-o(u )1Ar-o (7) )

A1-0 (8 )A1-o (8) 5

Mean

1 . 3 6 9 ( 1 )2 x 1 , 3 7 6 ( 1 )

L . 3 7 4

r . 9 7 r < 2 )2 x 2 . 0 1 8 ( 1 )

2 . L 3 7 < 2 )2 . 0 0 4 ( 1 )

2 . 0 2 6

1 . 9 9 5 ( 1 )1 . 8 9 3 ( r )1 . 9 5 5 ( 1 )1 . 9 0 0 ( 1 )1 . 9 3 2 ( 1 )1 . 9 0 0 ( 1 )

1 . 9 2 9

o (s ) r - o ( z )o(e) - o(e)1

Meen

o( r ) - o (z )?o ( 1 ) - o ( 6 ) :o ( 2 ) - o ( 2 ) zo t z l z - o ( o ) l0 ( 3 ) - 0 ( 2 ) .o ( : ) - o ( o ) 3o ( 6 ) - o ( 6 ) 3

llean

o ( 3 ) - 0 ( 6 )o ( : ) - o ( z ) s0 ( 3 ) - 0 ( 8 )o ( 3 ) - o 1 a ; s0 ( 6 ) - 0 ( B ) .0 ( 6 ) - 0 ( 8 ) ,0 (7)4- 0 (6) q0 ( 7 ) 4 - 0 ( 7 ) -o (7) -4 - o (8)0 ( 7 ) ) - 0 ( 8 )o ( 7 ) s - o ( 8 ) :0 (7)4- 0 (8) )

2 x 2 . 3 8 5 ( 1 ) 1 2 0 . 7 ( L )2 . 3 6 7 ( 3 ) 1 1 8 . 6 ( 1 )

2 . 3 7 9

2x2x

2x2x2x

2 . 6 7 3 ( 3 ) 8 4 . 1 5 ( 6 )3 . 0 4 7 ( 1 ) r 0 0 . 1 0 ( 6 )2 . 9 1 s ( 1 ) 9 2 . 4 s ( 6 )2 . 8 2 2 ( 2 ) 8 9 . 0 9 ( 5 )3 . 1 6 9 ( 2 ) 9 9 . 3 7 ( 6 )2 , 5 6 0 ( 2 ) 1 6 , 3 r ( 4 )2 . 8 1 1 ( 3 ) 8 9 . 0 i ( 8 )

2 . 8 5 6

2 . 5 6 0 ( 2 ) 8 2 . 3 3 ( 6 )2 . 8 8 8 ( 2 ) 9 s . 6 7 ( 6 )2 , 7 9 3 ( 2 ) 9 0 , 6 4 ( 6 )2 . 8 9 L ( 2 ) 9 s . 8 0 ( 7 )2 . 7 2 9 ( 2 ) 9 1 . 0 0 ( 5 )2 . 7 9 3 ( 2 ) 9 4 . 8 6 ( 6 )2 , 1 8 4 ( 2 ) 9 2 . 6 9 ( 6 )2 . 7 3 8 ( 1 ) 9 0 . s 0 ( 2 )2 . 8 9 8 ( 2 ) 7 7 . 6 4 ( 6 )2 , 4 1 7 ( r ) 9 6 . 4 r . ( 6 )2 . 8 2 5 ( 2 ) 7 8 . 2 O ( s )2 . 4 L 7 ( r ) 9 6 . 0 8 ( s )

r . r . -ocz)3,1 :* 2.536(2)na-o(a) i ' i 3x 2. i80(t )N a - o ( 5 ) - ' ' 3 x 2 . 7 0 4 ( 2 )

M e a n 2 . 6 7 3

H - 0 ( 3 ) 0 . 8 6 ( 4 )u.-o(s) 2.39(4)

Mean 2.728

op l l - ^01212;6^ t * 2 .s r4s( j ) 70 .17(6)0 ( 2 ) ; ' ; - o ( 4 ) ; ' ; 3 x 3 . o 7 a 4 ( 2 1 ) 7 0 . 6 I ( 4 )0 ( 2 ) l ' " - 0 ( 5 ) - ' - 6 x 3 . 5 8 6 s ( 2 3 ) 8 6 , 3 2 ( 4 to1+11-^oqt12q6^ l " 4 ,4576(6 , 106.6r (5 )0 ( 4 ) - ' " - 0 ( 5 ) - ' " 6 x 2 . s 6 6 5 ( 9 ) s 5 . 7 9 ( 4 )

Mean 3 .251

o(3) - o (5) 3 . r 9 L ( 2 ) 1 5 4 . ( 2 )

I = y - x , y , z i 2 = y - \ , - x , z i 3 = x , x - y , z ;5 - -y+2/3,x-y+1"/3,2+L/3i 6 = -y,r-y,z

coordinates to those of Table 4.

* = poslcloned ln an adjacent unit cel l .

4 - y-*I I 3,- \+2 / 3,2+2 / 3 it ransf omtions relat ing

(Barton, 1969), refinement of vanadium tourmalinein both orientations yielded variances and meanbond lengths which differed by insignifislalamounts.

Ditrigonality, 6, defined by Barton (1969) as ameasure of the distortion of the TuO,, ring from hex-agonal symmetry, is 0.009 for vanadium tourmaline.This is closer to the values for elbaite (0.009) andschorl (0.005) than to aluminous dravite (0.019), uv-ite (0.029), dravite (0.032), and buergerite (0.038).The only structural parameter which appears to becorrelated to ditrigonality is 9b octahedral angle vari-ance (Fig. 2). This apparent negative correlation isnot unexpected, because of vertex sharing betweenthe 9b octahedra and the Si tetrahedra of the ring.The reason for the gross deviation of buergerite fromthe general trend remains obscure.

FOIT AND ROSENBERG: VANADIUM-BEARING TOURMALINE

a 2Ue(tet)

(Deg?)

.U

ao

.A

av

OB

.E

os

l), adequately describes the site chemistry of van-adium tourmaline.

While very little substitution of cations other thanAl has been observed in the smaller l8c site (Tsang etal., l97l), the larger more distorted 9b site withwhich it shares an edge tolerates extensive as well asdiverse substitutions, producing a substantial varia-tion in mean bond length. The angular variance (dis-tortion) of both the 9b and l8c octahedra in tourma-line appears to be negatively correlated to mean 9b-O bond length, smaller size producing greater dis-tortion (Figs. 3, 4). However, no distinct correlationbetween the size of the l8c site and its angular vari-anoe was observed. The predominant influence of themean 9b-O bond length over the l8c-O bond lengthi11 s6lflelling l8c octahedral distortion is thought toreflect the significantly greater variability in the sizeof the 9b site. The larger angular variance in schorl(Fig. 3) than might be expected on the basis of 9b sitesize may be due, in part, to the Jahn-Teller effect(Walsh et al.,1974).

The volume of the unit cell varies non-linearlywith the dimensions of the 9b and l8c octahedra(Fig. 5); the greater the nean 9b-O and l8c-O bondlengths, the greater the cell volume. This accounts forthe success of the often-used Epprecht (1953)+ypediagram, by which cell parameters have been shownto vary as a function of substitution between schorl-dravite and schorl-elbaite. The relationships between6

2640 2 650 2 .660 2 .670

M E A N 3 a . O I N T E R A T O M I C

2 680 2 .690 2 7Ol

orsraNcg r ir

Fig. l. Variation of tetrahedral bond angle variance, da1,.,;, withmean 3a-O interatomic distance (A : aluminous dravite, B :buergerite, D : dravite, E : elbaite, S : schorl, U : uvite, V :vanadium tourmaline).

Octahedral portion. In vanadium tourmaline thereis a signifcantly greater scattering power associatedwilh the larger more distorted 9b site (Lo.urHvorr)than with the smal ler less distorted l8c si te(LonrHvo.or). This is in keeping with the results ofother transition-metal tourmaline refinements (buer-gerite, Barton, 1969; schorl, Fortier and Donnay,1975). A comparison of the mean l8c-O bond lengthand site oocupancy (1.926A, Lo"rHvo*) of vanadirmtourmaline to those of elbaite (1.9054, Al,.-) andschorl (1.922A, AlonrFe2o*o') is in accord with verylittle substitution of cations other than Al in thissite. Thus, the formula (Nao,*Cao.r"Mgo.,rKo.or)(Mg, ,rHv,.rrHv, ,r)(Al , ," Hvo oo)(Si, 63A10 rr)B, , ,O2ro,.,o(OHr.rrForr), which is very similar to the oneproposed on the basis of the analytical data (Table

aB

oo

au

Ao

au oE

oS

oo4o

o o35

qo30

o o 1 0

o 0 2 5

zo o.o200

o

60 65 70 ao a5 90 95 100

Oi,"o, ro.9"r

Fig. 2. Variation ofditrigonality, 6, (6 = (r1 - r)/r" where 11 andr" are tle distances from the three-fold axis to O(4) and O(5), re-spectively) with 9b octahedral angle variance, d4e6;.

OB

Eo

as.A

av

.D

au

FOIT AND ROSENBERG: VANADIT]M-BEARING TOURMALINE

a 5

/\2ve(9b)ao

( Deg.'?)

75

70

602 000 2 010 2 020 2 030 2 040 2 050 2 060

MEAN 9b-O INTERATOMIC DISTANCE ( i )

Fig. 3. Variation of 9b octahedral angle variance, &<gv>, with

mean 9b-O interatomic distance.

tourmaline cell volume and mean octahedral bondlengths and between mean 9b-O bond length andl8c octahedral distortion is not surpri5ing, in view ofthe fact that edge-sharing between the 9b and l8c oc-tahedra results in a quasi-rigid three-dimensional oc-tahedral framework.

Alkali cation site. The 3a site appears to be con-strained by the octahedral (9b) and tetrahedral por-tions of the structure, and therefore is only capable ofadjusting its size within restricted linits to variationsin cation occupancy. A plot of Na (largest abundantalkali cation in tourmaline) content vr. mean alkali-oxygen bond length (Fig. 6) reveals a positive corre-lation between sodium content and the size of the 3acoordination polyhedron. However, if mean 3a site-oxygen interatomic distance is plotted against eitherthe effective radius of the cations occupying that siteor cell volume (Fig. 5), no correlation can be ob-served, indicating that the octahedral and tetrahedralportions of the structure tend to hold the site openwhen cations smaller than Na occupy the site. Thisconclusion is in accord with the absence of a correla-tion between cell volume and 3a site occupancy ob-served in a study of tourmalines synthesized in thesystem MgO-AlrOr-SiO2-B2Or-HrO (Rosenberg andFoit, 1977), and demonstrates the relative in-significance of alkali-site occupancy as compared to

octahedral-site occupancy in determining cell param-eters.

Protons. The principal hydrogen bonds involvetwo non-equivalent oxygen atoms, O(l) with thebond direction paralleling the 3-fold axis and O(3)with the bond lying in the mirror plane. While it waspossible to refine the positional parameters of one ofthe protons successfully, revealing a hydrogen bondgeometry, H-O(3) : 0.86(4)A, H-O(5) : 2.39(4)Aand O(3)-H "' O(5) : 154(2)", comparable to thatreported for buergerite (Tippe and Hamilton,l97l),no information regarding the partial occupancy of ei-ther the O(l) or the O(3) proton site could be ob-tained from the X-ray results. An indirect estimationof proton occupancy was obtained, using the methodof Donnay and Allman-n (1970) (Table 6). Althoughthe discrepancy value Arms : 0.054, a measure of theagreement between calculated and ideal valencesumi- is comparable to those reported for othertourmalines; (Fortier and Donnay, 1975), the chargeimbalance for O(l) and O(3) is unusually large. Thissuggests that vanadium tourmaline has a slightlylower proton content than indicated by the analyticaldata, and that some of the V atd/or Fe is in a highervalence state. The oxidation state and site distribu-tion of vanadium in tourmaline from this locality iscurrently fsing studied using optical absorption tech-niques by G. Rossman (personal communication).

Bo

v, oo a

a

D

o U oas

51

ve(18c)

(Deg.2)

4 5

2010 2c20 2 030 2040 2 050 2 060

MEAN gb-o TNTERAToMTC otsrarcE (i)

Fig. 4. Variation of l8c octahedral angle variance, @re.; with

mean 9b-O interatomic distance.

3 9 L2 000

794

a 1 9 7 0

1 9 3 5

FO I T A N D RO S EN B E RG: I/A NA D I T] M - B EA RI N G TO U RMA LI N E

zS r.gos

o

=5 r soo

z1 955

o

I

S r 9 5 oOo

I 1,94s

oFIO l 9 4 O

3

Ao ,

x

seems to be remarkably tolerant of diverse sub-stitutions. Despite extensive dehydroxylation relativeto schorl, elbaite, and dravite according to the sub-stitution H* + R?;) : R?f,ou as well as substitution ofMg'* into the 9-coordinated 3a-"alkali" site (Foitand Rosenberg, 19771. Rosenberg and Foit, 1977), thestructural con-figuration of vanadium tourmaline ap-pears to be intermediate between these end-members(Figs. l-5).

Studies of both natural aluminous dravites (Table7, Foit and Rosenbery, 1977) and synthetic tourma-lines in the system MgO-A1rOr-SiOr-BrOr-HrO(Rosenberg and Foit, 1975,1977) suggest substantialAl3*-for-Mg'* substitution in the 9b octahedral site(Fig. 7), charge balance being maintained by dehy-droxylation, creation of alkali site defects, and/orAl'*-for-Sio* substitution. Similar extensive sub-stitutional relationships involving Al'* and Li* in the9b sites occur in natural elbaites (Table 7, Fig. 7).Synthesis of tourmaline in the system NarO-AtOr-SiOr-BrO3-HrO (Ekambaram, personal communica-tion) demonstrates that the octahedral sites may con-tain only Al. Thus, extensive and possibly completesubstitution of cations (Al3* for Mg2* and Li*), whichrepresent greatly diferent sizes ("'Li : 0.76, "Mg :O.'12, ut Al : 0.544; Shannon, 1976) and distortions(Figs. 2-4), takes place in the dravites and elbaites.

2.690

zoes ?

2.650 i;o

u

2675 0

c

z2.670 '

o

n

a.as II

2.66rJ

2655r540.o

Fig. 5. variati." ., *;:;";"*ith *"ight"d mean octahe-dral, I-gb-O + 2(TBc-O)l/3, and with mean 3a-O interatomic dis-tance. The mean 3a-O distance in uvite is 2.6,16A and lies belowthe abscissa.

Solid solution relationships: structural considerations

As observed in previous investigations (Buerger eral., 1962; Barton, 1969; Donnay and Bafion, 1972Fortier and Donnay,1975) the tourmaline structure

UFa

6

z

Ozfl

-ooz

o.60

o.50

o.40

o.302 640 2650 2.660 2.670 2.680

M E A N 3 A - O I N T E R A T O M I C O I S T N I \ I C E ( i )

Fig. 6. Variation of mean 3a-O interatomic distance with Na+ occupancy in the 3a ("alkali") site.

osB

o

AO

Eo

vO

oou

FOIT AND ROSENBERG: VANADIUM.BEARING TOURMALINE

Table 6. Vanadium tourmaline. Bond valenc€s (numb€r of bonds received by anion in parentheses)

795

la"lx Y z3a 9b 18c B S i

AnionoccuPancyf v-cAnion Mul t i

o ( 1 )

0 ( 2 )

o (3 )

o (4)

o ( 5 )

0 (6 )

o ( 7 )

0 .233( 1 )

0 . 1 3 1( 1 )

0 . 1 s 9( 1 )

0 . 4 3 3( 3 )

o .402( 2 )

0 .329( 1 )

1 . 000( 1 )

1 . 2 9 9

2 .O37

r , L 3 7

2 . 0 9 1

2 .057

1 . 9 8 4

2 ,027

0 . 0 3 7

0 . 093

0 . 091

0 . 0 5 7

0 . 0 1 6

o .o27

03.+r to. s , (0")0. zr o ' 111

0 .404( 2 )

02-

- 2 -(oH)0 . zzoo. z3

o-

t -o -

o -

o -

^z-u

18

18

0 . 4 1 1( 1 )

0 . 980( 2 )

0 .949( 2 )

1 . 014( 1 )

1 . 0 1 6( 1 )

0 ( 8 )

0 . 5 5 9( 1 )

0 . 4 6 4( 1 )

0 . 5 4 7( 1 )

u . ) 4 t( 1 )

0 . 4 9 8( r )

0 . 9 8 9( 1 )

2 .034 o. 034

A v = 0 . 0 5 4ms

L (Mean)

l ( M a x ) .

P G x P )v ( t )

2 . 6 7 3 2 . 0 2 63 . t 2 7 2 . 6 7 05 , 8 8 8 3 . 1 4 5L . 5 4 1 9 2 . 3 8 / 5

L . 9 2 9 r . 3 7 4 L , 6 2 52 .268 1 . 857 2 . 1385 . 690 2 .845 3 . 1683 . o r l 6 2 . 9 8 / 3 3 . 9 6 1 4

Structural configurations required by the nature andextent of coupled substitutions in natural as well assynthetic tourmalines appear to be easily attained,and are not markedly different from those previouslyreported.

The foregoing data have interesting implicationsrelative to the apparent lack ofextensive substitutionbetween dravite and elbaite. The tourmaline struc-ture contains two groups of crystallographically-dis-tinct octahedra; the edge-sharing l8c octahedrawhich spiral up the 3, screw axis and the edge-shar-ing 9b octahedra which cluster around the 3-foldaxis. These octahedra in turn share a common edge0(3)-0(6), forming a three-dimensional octahedralframework. Donnay and Barton (1972) have sug-gested that, due to the absence of variable valence-variable size cations (1.e., Fe2*'3*) along the dravite-elbaitejoin, the 9b and l8c octahedra cannot achievecon-figurations permitting a mutual sharing of edges,thereby making the formation of the tourmalinestructure impossible. However, no distinct couplingbetween the sizes of the 9b and l8c octahedra (Fig. 8)is observed in refned tourmaline structures. Further-

more, schorl and elbaite reflect a greater range of 9b-O bond length and l8c octahedral angle variancethan do dravite and elbaite (Fig. a). Yet natural Li-rich tourmalines containing appreciable amounts ofFe2* have been reported, whereas those containingan appreciable amount of Mg2* are unknown, eventhough these tourmalines may contain multiple-va-lent cations. If the availability of variable valencecations were an overriding factor lirriting the forma-tion of the tourmaline structure, despite possible ex-tensive disordering of cations between the 9b and l8csites, then natural and synthetic aluminous dravitesand natural aluminous elbaites would not form.Thus, there appears to be no obvious structural rea-son why dravite-elbaite solid solution should not bemore extensive than has been observed in naturalsamples.

Solid solution relationships: field strength consid-erations

Because of the flexibility of the tourmaline struc-ture and its tolerance ofsubstitutions in several sites,as well as the ease with which protons can be gained

796 FOIT AND ROSENBERG:

N a - A l T o U R M A L I N E

A13 al6

/ \

VA NA DI U M-BEARI NG TO URMA LI N E

and Li, both during petrogenesis and by the tourma-line structure as a result of the large diflerence in thefield strength of these cations. Less extreme fraction-ation between Fe'* and Li* due to the somewhatsmaller difference in field strengths would accountfor the greater abundance of elbaite-schorl solid so-lutions than elbaite-dravite solid solutions in nature(Foit and Rosenberg, 1977).

In the Brown Derby pegmatites, GunnisonCounty, Colorado, which are believed to have crys-tallized as a closed system from the walls inward, theatomic proportions of Mg2*, Fe'*, and Mn'?* (Staatzet al., 1955\ peak and diminish in turn during thecrystallization sequence, the first tourmalines beingrelatively Mg'*- and Fe'*-rich and the last being Li*-rich (Fig. 9). These data suggest that the highest feldstrength cations available were preferentially incor-porated in the tourmaline structure. The crystalliza-tion of l6urm4lins (and perhaps other minerals) re-sulted in the depletion ofhigh field strength cations,allowing Li* to enter the structure in progressivelygreater proportions during the later stages of crystal-lization. If cation field strength typically controls thecomposition of tourmaline to the same extent as inthe Brown Derby pegmatites, then solid solutions be-tween dravite and elbaite would be expected to bevery rare in nature.

Field strength can also be used to rationalize the

/ \

D R A V T T E ( s c h o r t )

M 9 3 a t uE L B A I T E

( L r 1 5 4 1 1 5 ) A t 6

Fig. 7. Extent (solid line) of octahedral substitution betweendravite, elbaite and Na-AI tourmaline in natural samples. Num-bers refer to samples listed in Table 7. Mg, Li, Al plus minor Fewere considered in plotting tourmaline compositions. Minoramounts of other elements were neglected.

or lost to maintain charge balance, the 9b octahedralsite will preferentially be occupied by cations withthe highest field strength (Z/a',where a : cation-an-ion interatomic distance; Weyl and Marboe, 1962).

The apparent absence of intermediates between el-baite and dravite in nature (Foit and Rosenberg,1977) may be due to the extreme fractionation of Mg

1 .

t

Table 7. Natural leumalins5 for which there appears to be significant Al3*-for-Mg2+ and -Li+ substitution in the 9b octahedral site(calculations based on 3l anions)

4 .

6 .

7 .

(*"0, 97t'0. 09*0. o, )

("tr. zrt"o. ,2tt0, o5A to.

or"to. ortro. or )AloBz . ss (tt,

. 9uoto. oo)oz zoo. 64 (o":

. l rto. o5 )

( " "o.57tro.3t l to .1t teo.or) (Msr.orA10.87F"o.03t io .oz)Ar6tz.ezs 'o.o4oz7oo.zo(o"a.z+to.os)

( t r0.67" 'o.z9c '0.04)(Msr.ooAlr . rot" i loat '310, %.or)AloBz.ossio.zsoztoo.o+(oHg.+o)

("80.47*t0.41K0. oz tr0. os ) (A1r

.52*s1 . 3stto. u""i loz )A1ou3. r3sio. goozzo3 . ea (o"0. ro )

(* '0 .7r t r ' .2 lc to.o7Ko.or) (Al t .er" i r .z7 Eo.o5tso.or t " i lor tno.or)Alous.oe(s is .osA1o.35)ozo.se(o"g.ssto.+z)

(* to.s6 Eo.35cto.08Ko.or) (A1r .z+" i r . .08 E0.o7Mto.o6)A1ot3.oa( t ts .gsolo.os)ozzoo.zs(o" t .zo.o.ss)

(*"0.6g t ro.14ct0.12*o.os)(Arr .ggl io .g4wo.rz t ro.oe" t3 lo: t " i lor t to .or)ot 6tz .a46Ls.azAlo. te)ozzoo.rs(o" : .os)

(* t1.o4t to. ro*o.ro)(ot r . r r " io .59tro.zzt t3 l r r " "o.oo" t o.or)o ' r t r .n6( t t r . ruolo. ro)orror .or@'r .gsFo.os)

1 . Swanson e t a1 . ( 1964 )2 . S lnpson (1931 )3. Takeehl (1953)4 . Chebo ta rev e t a l . ( 1971 )

5. El-Hinnar,r i et a l . (1966)6 . E l -H innaw i e t a l . ( 1966 )7. Chaudhry and Howie (1976)8 . S j i i g ren (1916 )

8 .

following observations regarding the extent of solidsolution in natural tourmalines between the end-members: schorl-dravite, Na(Fe,Mg)rAL(BO3),SLO,,(OH)o; elbaite, Na(Li,Al).AL(BO,),Si.O,.(OHL; and R*RI*R:*(BO')'Si.O"O3(OH), whereR* and R'* are principally Na* and Al'*, respectively(see Fig. 9 in Foit and Rosenberg,1977).

(1) While a few samples approximating end-member schorl/dravite have been observed, el-baites closely approaching end-member composi-tion have not been reported; all are appreciably Li-deficient and Al-rich. Although liddicoatite, theCa-analog of elbaite (Dunn et al., 1977) has a Li/Al ratio highsl than that of elbaite, it is also Li-deficient with respect to its idealized end-membercomposition.

(2) Most tourmalines which approach towithin 90 mole percent of end-member schorl/dra-vite are Mg-rich.

(3) Substitution of R3* for Mg'*, Fe'* or Li* inthe 9b site is extremely cornmon and generallymore extensive in the elbaites than in the schorlsand dravites.

'197

so

Do

au

vo

AoE

o

ao

1 900 1 9 O 5 1 9 1 O 1 9 1 5 1 g 2 o 1 9 2 5 1 ' 9 3 0 1 ' 9 3 5

M E A N 1 A c - O I N T E R A T O M I C O t S r l H c r ( i )

FOIT AND ROSENBERG: VANADIUM.BEARING TOURMALINE

.3

Uzfo

o

z

9

z

2.060

2 050

2.O40

2.O30

2 020

2.O10

2 000

Fig. 8. Variation of mean 9b-O interatomic distance with mean

I 8c-O interatomic distance.

(4) Li substitution in schorls and dravites isfar more limited than Fe and Mg substitution in el-baites.The preferential incorporation of higher field

EIgl -El - z

v\w

u . z

o.3

o.4

o.6

t\ '7

o.o o.1 02 0.3 0.4 0.5 0.6' o'7 0.8

ATOM FRACTIONL i . M g + F e + M n

Fig. 9. Variations in the Mg-, Fe2*-, Mn2*-, and Li-content of tourmalines from the Brown Derby pegmatites; calculated from the

data of Staatz et al.(1955). Squares, Mg/(Li+ Mg + Fe + Mn); circles, Fe2*/(Li + Mg + Fe + Mn); trianglcs, i[.dn2+/(Li + Mg + Fe +

Mn). Open symbols, Brown Derby No. 2 pegrnatite, filled symobols, Brown Derby No. 3 pegmatite. Arrows show direction of change

during crystallization from the walls inward toward the core. (*), Mg value used is a minimum.

za)

OtrtL

F

t/

L i

798 FOIT AND ROSENBERG: VANADIUM-BEARING TOURMALINE

strength cations (Al3*, Fe3*) irnplied by these rela-tionships is facilitated by the apparent ease of protonexchange which serves to maintain charge balance.

Although diverse substitutions in the tourmalinestructure do result in signifisalt systematic changesin polyhedral distortion (Figs. l-4), distortion itselfdoes not appear to play a major role in limiting sub-stitution. The nature and extent of cation substitutionin tourmaline appears to be more closely governedby fractionation resulting from differences in cationfield strength. Substitution of high field strength cat-ions generally results in improved local charge bal-ance (Foit and Rosenberg,1977) which undoubtedlyfacilitates these substitutions.

AcknowledgmentsThe authors are indebted to Dr. Kenneth G. Snetsinger, Space

Sciences Division, NASA Ames Research Cent€r. for informationregarding the location of the V-tourmaline deposit; Mr. DavidPlant, Department of Geology, Manchester University, for theelectron microprobe analysis, Professor Roger D. Willett, Depart-ment of Chemistry, Washington State University, for the use of hisautomated single-crystal diffractometer, and Ms. Julie V. Jaquishfor collection and reduction of the X-ray data. We also thank pro-fessor Wayne A. Dollase, Department of Earth and Space Sci-ences, University of California, Los Angeles, and professor Ga-briclle Donnay, Department of Geological Sciences, McGillUniversity, for critically reviewing the rnanuscript.

ReferencesBarton, Jr., R. (1969) Rcfinement of the crystal structure of

buergerite and the absolute orientation of tourmatines. ;lcraCrystallogr., 825, 1524-1533.

Buerger, M. J., C. W. Burnham and D. R. peacor (1962) Assess_ment of several structurcs proposed for tourmaline. Acta Crvs_tallogr.,1J, 583-590.

Busing, W. R., K. O. Martin and H. A. Levy (1962) A Fortratrcrystallographic least squares program. II.S. Nat. Tech. Inform.Serv., ORNL-TM-305.

--, - and - (1964) A Fortran crystallographicfunction and error program. Il.S. Nat. Tech. Inform. Sen.,ORNL-TM-306.

Chaudhry, M. N. and R. A. Howie (1926) Lithium tourmalinesfrom the Meldon aplite, Devonshire, England. Mineral. Mag.,40,747-75t.

Chebotarev, G. M. and G. P. Chebotareva (1971) Dravite varietyof tourmalinc from Muruntau (western Uzbekistan). Zap. llzbe-kistansk. Otd. Yses. Mineral. Obshchest.,24, ll2-1I5.

Cromer, D. T. and J. T. Waber (1965) Scattering factors computedfrom relativistic Dirac-Slater wave functions. Acta Crystallogr.,18. 104-109.

Donnay, G. and R. Allrnenn (1920) How to recognizs Or-, (OH)-and H2O in crystal structures determined by X-rays. Am. Min-eral., 55, 1003- 1015.

- and R. Barton, Jt. (19'12) Refinement of the crystal struc_ture of elbaite and the mechanism sf leurmatins solii solution.Tschermaks M ineral. P et ro g. M it t., I B, 27 3-296.

Dunn, P. J., D. E. Appleman and J. E. Nelen (1977) Liddicoatite,a new calcium end-member of the tourmaline group. Am. Min_eral.,62, ll2l-1124.

El-Hinnawi, E. E. and R. Hofmann (1966) Optical and chemicalinvestigation of nine tourmalines (elbaite). Neues lahrb. Min-eral. Monatsh,SO-88.

Epprecht, W. (1953) Die Gitterkonstanten der Turmalin. Schweiz.Mineral. Petrog. Miu., JJ, 481-505.

Foit, Jr., F. F. and P. E. Rosenberg (1974) Coupled substitutionsin tourmaline (abstr.). EOS, Trans. Am. Geophys. (Inion, 55,467.

- and - (1975) Aluminobuergerite, Na,*,Rl*R;+(BO3)3Si6Or8O3_,(OH)r+., a new end-member of thetourmaline group (abstr.). EOS, Trans. Am. Geophys. Union, 56,46t.

- and - (1977) Coupled substitutions in the tourmalinegroup. Contrib. Mineral. Petrol.,62, lW-127.

Fortier, S. and G. Donnay (1975) Schorl refinement showing com-positional dependence of the tourmaline structure. Can. Min-eral., I 3, 173-177 .

Robinson, K., c. V. Gibbs and P. H. Ribbe (1971) Quadraticelongation: a quantitative measure of distortion in coordinationpolyhedra. Science, I 72, 567 -570.

Rosenberg, P. E. and F. F. Foit, Jr. (1975) Alkali-free tourmalinesin the system MgO-AI2O3-SiO2-H2O-B2O3 (absrr.). Geol. Soc.Am. Abstracts with Programs, T,1250.

- and - (1977) Crystal chemistry of alkali-free tourma-line (abstr.). Geol. Soc. Am. Abstracts with Programs, 9, ll47-I 148.

Schemtzer, K. (1978) Vanadium III als Farbtrdger bei naturlichenSilicaten und Oxiden-ein Beitrag zur Kristallchemie des Van-adiums. Doctoral Dissertation, Ruprecht-Karl-Universitiit, Hei-delberg.

Shannon, R. D. (1976) Revised effective ionic radii and systematicstudies of interatomic distances in halides and chalcogenides.Acta Crystallogr., A 3 2, 7 5 l-7 67.

Simpson, E. S. (1931) Contributions to the mineralogy of WesternAustralia. lY. J. Roy. Soc. W. Austrulia, 17, 137-149.

Sj<igreq H. (1916) The chemical composition of tourmaline fromUt6. Bull. Geol. Inst. Univ. Upsala, 15,317-324.

Snetsinger, K. G. (1966) Barium-vanadium muscovite and van-adium tourmaline from Mariposa County, California. Am. Min-eral., 5I, 1623-1639.

Staatz, M. H., K. J. Murata and J. J. Glass (1955) Variation ofcomposition and physical properties of tourmaline with its posi-tion in the pegmatite. Am. Mineral.,40,789-804.

Swanson, H. E., M. C. Morris, E. H. Evans and L. Verner (1964)Standard X-ray diffraction patterns. U.S. Nat. Bureau of Stan-dards, Mem 25, Part I; Sec. 3,4748.

Takeshi, H. (1953) White tourmaline from Kanakura Mine, Na-gano Prefecture, Japan. J. Mineral. (Japan), I,105-112.

Tippe, A. and W. C. Hamilton (1971) A neutron-difraction srudyof the ferric tourmaline, buergerite. lm. Mineral., J4 l0l-l13.

Tsang, T., A. N. Thorpe and G. Donnay (1971) Magnetic suscepti-bility and triangular exchange coupling in the tourmaline min-eral group. f. Phys. Chem. Solids,32, l44l-14ts,8.

Walsh, D., G. Donnay and J. D. H. Donnay (1974) Jabn-Teller ef-fects in ferromagnesian minerals: pyroxenes and olivines. gzlLSoc. fr. Mineral. Crystallogr.,97, l7Ll83.

Weyl, W. A. and E. C. Marboe (196\ fhe Constitution of Glasses:a Dynmic Interpretdtion. Interscience-Wiley, New york.

Manuscript received, September 26, 1978;accepledfor publication, February 28, 1979.