American Mineralogist, Volume63, pages1053-1073, 1978

Newbiopyriboles from Chester,Vermont: II. Thecrystal chemistry ofjimthompsonite, clinojimthompsonite,and , and the - reaction

DevIo R. Vnsr-nNleNn CHnRlnsW. BunNHepr Department of Geological Sciences,Haruard Uniuersity C ambridge, M assachus e t ts 02 I 38

Abstract

The crystalstructures of ,clinojimthompsonite, and chesteritehave been solvedand refinedusing three-dimensional X-ray intensitydata. Jimthompsonite and clino- jimthompsonitecontain I-beams composed of triplesilicate chains and wide octahedral strips. In jimthompsonitethese l-beams are stackedlike those of orthopyroxenesand ortho- ,while the l-beamsof clinojimthompsoniteare stacked like thosein clinopyrox- enesand clinoamphiboles. Chesterite contains both double- and triple-chain I-beams alternat- ing in theb directionand assembled in theorthopyroxene-orthoamphibole stacking sequence. The unnamedmineral that occursas lamellaein chesteriteis presumablythe monoclinic analogof this structure. The new mineralsare very similar to other low-calciumpyriboles with respectto several crystal-chemicalattributes, including polytypism, chain rotation,chain warping,lamellar orientation,and cationordering patterns. Examination of theoutermost polyhedra of chain silicateI-beams suggests that A-chain rotationsin orthorhombicpyriboles are a necessary consequenceof edge-sharingbetween tetrahedra and octahedra. As intermediatereaction products,the new structuresdelineate 4 reactionpath from anthophylliteand cummingtoniteto .This amphibole-to-micamechanism is not simple, but ratherproceeds by reconstructionofdouble-chain I-beams to formalternating double and triplechains, pure triple chains, and finallycontinuous sheets of silicatetetrahedra.

Introduction thompsonite,and chesteritewill then be presentedin Of the many new mineralsdescribed each year, turn. Comparisonsof biopyribole bond distancesand mostare not silicates,and thosethat areusually bear angles, stacking, exsolution directions, chain warp- no clearrelationship to the importantrock-forming ing, cation ordering, and chain rotations will follow. mineralgroups. The fortuitousdiscovery of four new Particular attention is given to chain rotations in mineralsthat are intimatelyrelated to pyroxenes, orthorhombic pyriboles, becausethis topic has re- amphiboles,and is thusunusual (Veblen, 1976; ceived so much attention in recent years and remains Veblen and Burnham, 1975,1976; Veblen et al., a point of controversy. In the final section of this 1977),and the insightsinto biopyribolecrystal chem- paper we discussamphibole-mica reactionsin light istry that theyprovide are profound. A previouspa- of the new minerals. perof this seriesdescribed physical the andchemical Model derivations characteristicsof thenew minerals (Veblen and Burn- ham, 1978).This paperdescribes their crystalstruc- Idealizedpyribole l-beamspossess either mirror or turesand presentsa "comparativeanatomy" of the c-glidesymmetry. Those containing chains that could low-calciumbiopyriboles. be assembled from an even number of pyroxene We beginby describingthe developmentof struc- chains("even chains") are bisectedby mirrors paral- tural modelsfor the new biopyriboles.The crystal lel to (010), while those with chains that could be structurerefinements of jimthompsonite,clinojim- assembledfrom an odd number of pyroxene chains ("odd chains") are cut by c-glidesparallel to (010). I Present address: Departments of Geology and Chemistry, Both even- and odd-chained ideal I-beams further Arizona State University, Tempe, Arizona 85281. contain 2-fold axes passing through the M-sites and 0003404x/78/l r r2-1053$02.00 1053 1054 VEBLEN AND BURNHAM: BIOPYRIBOLES

Table l. Crystal data and intensity measurementconditions Burnham, 1978,Table l). The similar intensitydistri- butions on O-levelb-axis precessionphotographs of Cl1no Jinthofrpsonlte j imthoEpsonite* Chesteri te chesterite,, and enstatite (Fig. I ) in- dicate that in projection down b thesestructures are

Maxlnun crysEal 0.10x0.11x0.52 0.05x0.08x0-40 0.06x0- 10x0.35 closely related. On this basis alone the new mineral was assumedto be a pyribole. Analogous reasoning volume,** 0.0044 Crystal 0.0011 0.0013 was used by Warren (1929) and Warren and Modell Llnear absorption 21.0 2l.O 20.5 (1930) to solve the first amphibole structures from coefflcient, cm-1 their relations to the pyroxenes. Crystal axis for c c c data collectlon Of the spacegroups consistentwith the chesterite 20 scan speed lo/nln. Lo/rir. r /m1n. diffraction symbol (A21ma, Am2a, Amma), only Total backgtou.d 40 sec. 80 sec. 40 sec. countlng tine A2rma is acceptableas pyribole symmetry; mirrors parallel to (100) are inconsistent with pyribole I- * Clinojinthonpsonite crystal data refer Lo the entire beam topology and linkage operations. The metric fragment. Diffraction was fr@ two lanellae (about 102 of the f ragnent) in jinthompsonlte. requirements and l-centering of the cell, and the ** calculated by absorption correction program, taking account of presenceparallel to (010) of both mirrors and c-glide planes all bounding crystal, planes in the chosen space group, led to the con- clusion that the chesteritestructure must consistof parallel to the b-axis,but this rotational symmetry is both even- and odd-chainedl-beams, alternating in destroyed if the two chains of an I-beam differ in the 6 direction. There were two possibilities:mixed some way. Ideal odd-chained l-beams (triple, quin- single and quadruple chains or mixed double and tuple, etc.) are thus symmetricallyequivalent to ideal triple chains.The latter was consideredmore likely. pyroxene I-beams, while even-chained I-beams Soon after the development of the chesterite (quadrupfe, sextuple, etc.) have the same linkage model, jimthompsonite was discovered.With b = characteristicsas amphibole l-beams. These consid- 27A (Veblen and Burnham, 1978,Table I ) and space erations were instrumental in the formulation of group Pbca (the same as orthopyroxene), it was clear structural models for the Chesterpyriboles. that if the reasoningbehind the chesteritemodel was The first new mineral to be recognizedwas chester- correct, this mineral should consistof triple-chainI- ite. Like the orthopyroxenesand orthoamphiboles, beams arranged in the same fashion as the single- it was observedto have a= l8 and c= 5VqA,and its chain I-beamsof orthopyroxene.Like chesterite,jim- 45A b-axis is an integral multiple of 9A (Veblen and thompsonite closelymatched the orthopyribole (iOl) diffraction intensity distribution (Fig. l). Clinojimthompsonite and the unnamed mineral a* were later discoveredas lamellaein jimthompsonite and chesterite.The fact that their a dimensionsand B angles(Veblen and Burnham, 1978,Table I ) are close to those of low-calcium clinopyroxenesimmediately suggestedthat they are monoclinic polytypescontain- ing triple chains and mixed double and triple chains respectively.

Experimental The nearly colorlesssingle crystals used for X-ray intensity measurementare elongated in the c direc- tion, as a result of their excellentprismatic cleavages. Approximate dimensionsof the crystalsare listed in Table l. Becausethe Chester pyriboles are inter- grown with each other, jt is not easyto obtain mono- Fig. l. O-levelD-axis precession photographs ofenstatite (En), phase crystals of a size suitable for data collection. anthophyllite(An), jimthompsonite (Jt), and chesterite(Ch). The similarities in intensity distribution between the four minerals The chesteritecrystal appeared to be truly mono- first led to the conclusion that chesterite and jimthompsonite are phase, but was small. The jimthompsonite crystal pyriboles. The vertical direction is a*, and c* is horizontal. gave extremely weak diffractions from anthophyllite VEBLEN AND BURNHAM: BIOPYRIBOLES 1055 and chesterite,and precessionphotographs showed minimumobservable intensity criteria were the same light streaks in the b direction, probably indicating as thosedescribed by Burnhamet al. (1971),except some chain-width disorder. During intensity mea- that the minimumobservable intensity for clinojim- surement, these streaks interfeied with (ftkO) reflec- thompsonitewas setat 3o7,rather than 2ot. Unob- tions with h/2 + k : 2n * l, which were observedto served reflections were excluded from the least- be systematicallyabsent on precessionfllms. These squaresnormal equations matrices in all cases.Initial structure factors were therefore removed from the scalefactors were obtained from a Wilson plot pro- data set, and structure factor calculationsusing the gram(Hanscom, 1973). refined structure later showed that none of theseF"s The three structureswere refined with the full- have a magnitude above the minimum observable matrix least-squaresprogram RnlNr (Finger,1969), level. The data for clinojimthompsonite were col- usingscattering factors given by Cromerand Mann lected from two (100) lamellae in a crystal of jim- (1968)for Mg2+,t"'+, Si'+, and O1-; anomalous thompsonite. Each lamella was about 2p wide, and dispersioncorrections from Cromer and Liberman they were in the same crystallographicorientation. (1970)were applied to the scatteringcurves during The diffractionsfrom this crystalwere sharp, with no the finalstages of therefinements. The small amounts evidenceof streaking. of Ca'+, Al3+, and Mn2* were ignoredin the site Intensity data were measuredusing a Picker Fe,cs- occupancyrefinements. I four-circle diffractometer and Nb-filtered MoKa radiation. They were correctedfor Lorentz, polariza- Jimthompsonite tion, and absorptioneffects as describedby Burnham et al. (1971). Scan speedsand background counting Model deriuation times are listed in Table l. A model-dependentlinear Initial atomic coordinatesfor a trial model of jim- absorption coefficientwas derived for jimthompson- thompsonite were derived by cutting the ideal, unro- ite, because density and complete chemical data tated orthoenstatitestructure along its c-glide planes were not available.An electronmicroprobe analysis and inserting additional silicate tetrahedra,octahe- and an assumed cell content yielded cation ratios dral cations,and OH'- ions in a mica-like configura- which, when combined with the cell volume, permit- tion to form triple-chain I-beams.The I-beam stack- ted calculation of the linear absorption coefficient. ing sequence and inter-I-beam linkages of the The same value was assumedfor clinojimthompson- pyroxene were thus preservedin the model. ite, and an absorption coefficient was calculated for Nomenclature for the 29 atoms in the asymmetric chesterite,based again on assumedcell contents.Sub- unit is shown in Figures 2,3 and 4. Octahedral cat- sequent refinement of the structures demonstrated ions (M), Si, and O are numberedby startingat the c- the validity of these absorption coefficient calcu- glide and counting outward to the edgeof the I-beam, lations.The linear absorptioncoefficients used for the in the b direction; the sitesanalogous to those con- absorption correctionsare listed in Table l. taining OH'- or Fr- in the amphibolesand micas are Weighting of observationsduring refinementand designated "OH." The Si and O atom labels are

Fig 2. Triple A-chain of jimthompsonite,projected onto (100). The inner parts of the chain are very straight, while the outer parts are slightly O-rotated. The c-glide is labelled"9", and M5 is indicated by solid circles. 1056 VEBLEN AND BURNHAM: BIOPYRIBOLES

'L_

Fig. 3. Projection onto (100) of the O-rotated triple B-chain in jimthompsonite. The B-chain clearly is rotated more than the A-chain. appended by A or B to indicate the tetrahedral chain the B-chain, where 068 is closer to the c-glide than to which they belong; OH atom labels are similarly o58. appendedto indicate on which side of the octahedral bandsthey are located.As in the orthopyroxenesand Structure refinement orthoamphiboles,the A-chain is cut by a b-glidepar- The initial scalefactor was refinedfor one cycle, allel to (100) and lies between octahedral layers of resultingin R*","n,"o(Rr) : 48.0 percentfor the opposite skew, while the B-chainis piercedby 21axes starting model. After five least-squarescycles in and lies between octahedral layers of like skew (see which the scalefactor and atomiccoordinates were Thompson, 1970, for definition of skew). The only allowed to vary, R reached8.4 percentfor 2800re- exceptionto theserules for atom nomenclatureis in flections.The scalefactor, atom positions,and iso-

"l--

Fig.4. Jimthompsonitestructureprojectedonto(001).Thestackingsequenceis**--.B-chainsoccurbetwsenoctahedrallayersof like skew, while A-chainsare betweenlayers ofopposite skew.The basaloxygen layers of the silicatechains are warped out of the (100) plane.The c-glideand D-glideare labelled"g" and "gr." N,t5is indicated by solid circles,and the unit cell is outlined. VEBLEN AND BURNHAM: BIOPYRIBOLES 1057

Table 2. Discrepancyfactors, R, for jimthompsonite, Table 3. Jimthompsonite atom coordinates, temperature factors, clinojimthompsonite,and chesterite and metal (M) site occuPancies

At on Fe/ Felug Clino- Jlnthonpsonite Jimthmpsonlte Chesterite l{1 0.1255s(11)0.27883(6) 0.89653(35)0. 43(4) 0.050(6) M2 0.12535(r0) 0.33s33(5) 0.39566(32)0.58(4) 0. rs6(6) M3 o.12506(t2) o,392s7(6) 0.89517(36) 0.42(4) 0.078(5) !t4 o , r24a5(12) o.4s165(7) 0.39642(37)0.3s(4) 0.011(6) A11 observations: M5 o. 12369(6) o. so690(4) 0.89507(19)0.55(2) 0.803(7) Unweighted R* 0.121 0. 181 SiIA 0.2711r(9) 0,27787(7) o.ss957(27) 0.30(3) tJeighted R 0.068 0.056 Si2A o.26964(rO) 0.39005(5) 0.55980(30)o.4C(3) Nunber of obs. 6815 5043 Si3A o.21242(9) 0,4470C(6) 0.06669(29) O.42(3) 0.48025(9) o.27769<1) 0.26700(28)0.33(3) Observed dif f ractlons : si2B 0. 48078(10) o .38929(6) 0.27218<3o) 0.43(3) Uoweighted R 0.086 0. 084 0.091 si3B 0.47s79(10)0.44553(7) 0.77466(30) 0.47(3) Weighted R 0.058 0.075 0. 059 oliA 0.18329(22) 0.33519(16)o.0624r(69) D,49(6) Number of obs, 4491 169 2633 OHB 0.5686r(22) 0.33s13(17)0.76931(69)0.s0(6) otA 0.30061(23)0.24981(15)0.31159(73)0.65(7) Unrejected observatlons : c2a. 0.18399(22)0,21777 (16) o.s6r40(59) 0.40(6) Unweighted R 0.084 0. 067 0.081 o3A o , 29949 (22) 0.33401(r7) 0.s5638(70) 0.68(6) Welghted R 0.057 0.055 0.045 04A 0.r8325(23) 0.39170(1s)0.56291(74)0.50(7) Number of obs. 4457 747 2555 05A 0.29980(23) 0.4r49r(r6) 0.82585(78)0.73(7) 06A 0. 30187 ( 23) o.42028(16) 0.33r48(77) 0.69(7) Relection criterion: AF>20.0 AF>20.0 Ar>35.0 o7A 0. r8560(22) 0.44852(16) 0.06801(74)o.ss(6) o8A 0.3r330(22) 0.49841(15) 0.08080(77)o.6r(7) olB 0.4s141(24)0,24443(15) 0.03443(70)0.54(6) * unwetshredn = rllroo.l-lr""rll/IlFob..l 02B 0,56718(22) o.21792<16)0.26828(69) 0.4r(6) o38 o .45168(22) 0.33338(r7) 0.23045(70) o. 6s(6) t.Jeighted R = [E w(FoO"-F".r)2/I (15) O.60(7) "rzo.]! 04B 0. s6725(24) 0.39207(15)o.27127 o58 0.44881(23)0.4244r(15) 0.05r68(74) 0.59(6) 069 0.45r38(23) 0.40886(16)o.s4386(77) 0.76(7) olB o .56296(23) 0.44849 ( 16) 0.177r8<74) O.63(1) tropic temperature factors were then refined for three o88 o.43394(23) 0.49561(16) o.7r39r(7s) 0,62(7) cycles. Anomalous dispersion corrections were in- troduced, and the scale factor, atomic coordinates, Table 5'?,and observedand calculatedstructure fac- isotropic temperaturefactors, and unconstrainedM- tors are given in Table 8. site occupancieswere refinedusing the completedata set. Final R values are given in Table 2. Atomic 'To coordinates, isotropic temperature factors, and M- obtain a copy ofTable 5 (bond distancesand angles),order Document AM-78-083, and to obtain Table 8 (structurefactors), are listed in Table 3, and selected site occupancies order Document AM-78-084, from the BusinessOffice, Mineral- bond distances and angles appear in Table 4a. A ogical Society of America, 1909K Street,NW, Washington, DC more complete listing of distancesand angles is in 20006 Pleaseremit $1.00 in advance for each microfiche.

Table 4a. Selectedbond distancesand anglesin jimthompsonite

o M-0 Distances. I Si-O Distances, A

M1 octahedron M2 octahedron M3 octahedron SiLA Tetrahedron Si2A Tetrahedron Si3A Tetrahedron M1-02A 2.082 M2-0HB 2.066 M3-0HA 2.097 silA-olA 1.6L6 si2A-o3A I.624 Si3A-05A L.628 M1-02A(1) 2.078 M2-O2A 2.L02 M3-0HB 2.078 sirA-o1A(1) r.628 si2A-04A 1.610 513A-06A L.673 u1-o2B(2) 2.082 M2-OHA 2.069 M3-O4A 2.070 sllA-o2A L.623 si2A-o5A L.6L7 S13A-O7A 1. 618 M1-028(3) 2.083 M2-O4A 2.075 M3-07A 2. T04 silA-03A 1.619 si2A-064 r.622 S13A-08A 1.595 M1-0HA 2.069 viz-o2B 2.09r M3-o4A 2.066 M1-0nB 2.069 t'12-o4B 2.083 M3-078 2.I20 Mean l.62L Mean 1.618 Mean r.628

Mean 2.077 Mean 2.081 Mean 2.089 S11B Tetrahedron Si2B Tetrahedron Si3B Tetrahedron stlB-o1B r.62L Si2B-O3B t.632 Si3B-O5B L.654 M4 Octdhedron M5 0ctahedron s118-018(1) 1.630 S12B-04B L.612 Si3B-068 1.643 M4-04A 2.151 M5-O7A 2.167 M5-O5B 2.938 sirB-o2B 1.619 Si2B-O5B 1.625 Si3B-07B I.626 M4-04B 2.r38 M5-O7B 2.r54 si3A-M5 2.889 silB-o3B 1.619 Si28-068 L.527 Si3B-08B r.604 M4-078 2.080 M5-08A 2.042 M4-O7A 2.077 M5-O8B 2.OOZ Mean 7.622 Mean 1.624 Mean I. OJI M4-O8A 2.032 M5-06A 2.443 o M4-088 2.048 M5-06B 2.800 Estinated Standard Error Si-O = 0.005A

Mean 2,087 Mean 2.268 Tetrahedral Chains

Estimated Standard Error M-O = 0,0053 A Chain B Chain 0rA-o1A-o1A r7g.60 o1B-o1B-O1B 166.90 05A-06A-05A 173.50 o58-068-O5B I6L,70 Estimated Standard Error O-0-O = 0.40 1058 VEBLEN AND BURNHAM: BIOPYRIBOLES

Table 4b. Selectedbond distancesand anglesin clinojimthompsonite

o o M-O Distances, A Si-O Distances, A

Ml Octahedron M2 Octahedron M3 Octahedron Sil Tetrahedron Si2 Tetrahedron Si3 Tetrahedron 2 MI-OH 2.06L 2 M2-oH 2.079 2 M3-OH 2.098 sil_ol L.628 s12_03 r.606 si3_o5 t.642 2 M1-O2 2.072 2 M2-O2 2.086 2 M3-O4 2.073 sil-ol(5) L.627 Si2-O4 L.6L3 Si3-06 1.655 2 M1-o2(4) 2.O7a 2 M2-O4 2.081 2 rr3-O7 2.091 sil-o2 r.534 si2-05 L.628 si3-o7 r.632 si1-o3 1.616 Si2-05 r.642 Si3-O8 1.501 Me an ?.010 2.082 Mean 2.089 r.626 L.622 1. 633

M4 Octahedron M5 Octahedron o M4-O4 2.L28 2 vr5-o7 2.L62 M5-O5 3.106 Eatimated Standard Error Si-O = 0.0094 M4-O7 2.081 2 M5-O8 2.032 si3-M5 2.979 M4-08 2.O49 2 M5-O6 2.659 Tetrahedral Chain

Meah 2.086 Mean 2.284 o1-o1-o1 17r.40 +0.6d 05-06-05 170.00+ 0.60 o Estimated Standard Error M-0 = 0.0094

The jimthompsonite structure axis projectionof the structure(Fig. 4). The basal Projectionsof the refinedjimthompsonite structure facesof the Si3,{ and Si3Btetrahedra are tilted signif- are shown in Figures2,3, and 4. I-beams consisting icantlyout of the (100)plane. The structuralsignifi- of triple silicatechains and wide octahedralstrips are canceof pyribolechain warping is discussedin a later connectedin the orthopyroxene-orthoamphibole * section. *-- stackingsequence (see Veblen et ql.,1977, for an explanation of stacking terminology). Each sili- Clinojimthompsonite cate chaifr has two symmetricallyindependent chain rotation angles,an inner rotation angle(Ol-Ol-Ol) Model deriuations and an outer rotation angle (05-06-05), unlike the As shownoriginally by Warrenand Modell (1930), pyroxenes and amphiboles, which have only one. the orthopyroxeneunit cell can be closelyapproxi- Both chains are O-rotated, the A-chain with inner matedby two unit cellsof clinopyroxenerelated by a and outer rotation anglesof 179.6"and 173.5o,and 6-glideparallel to (100).Since an analogousrelation- the B-chain with inner and outer anglesof 166.9'and shipwas hypothesized for jimthompsoniteand clino- 161.7". Chain rotation angles in pyriboles are dis- jimthompsonite,the sametransformation that Smith cussedlater in this paper, in light of the new struc- (1969)used to relatelow clinoenstatiteatom coordi- tures; Thompson (1970) explains O-rotation us. ^S- natesto thoseof orthoenstatitewas used to derivethe rotations. The silicate chains are topologically dis- clinojimthompsonitestarting coordinates. The fol- tinct from the triple chains in BarSirOr, in which lowingoperation was applied to atomsin one asym- alternating apical oxygens point in opposite direc- metricunit of the refinedjimthompsonite structure: tions (Katscher and Liebau, 1965;Liebau, 1972\. xcn: 2(xy- l/4) As in pyroxenesand amphiboles,the outer M-site is a distorted octahedron (M2 in pyroxenes,M4 in lctt: lJt 2"11= l/2 I zy * - l/2)a11 tan(B"p - 90")/cy amphiboles, and M5 in jimthompsonite), while the @y inner M-sites are more regular octahedra, slightly The resultinEP2r/c model is a triple-chain analogof compressedin the [00] direction. By further analogy the low clinoenstatitestructure. to the low-Ca pyroxenes and amphiboles, Fe2+ is Because the observed space group of clinojim- concentratedin the outer, distorted M-site, while the thompsonite is C2/c, the P2r/c model was modified regular M-sites favor Mg. The refinedFel(Fe + Mg) by an origin shift and by rotation of the B-chain ratio of 0.22 agreeswith the (Fe + Mn)/(Fe t Mn * atoms and OHB ion about the C2/c 2-fold axis.The Mg + Ca) value of 0.22 which was measured for this positions of corresponding atoms in the A- and B- crystal by electron microprobe analysis. chains were then averaged. The structure described Chain warping or corrugation is apparent in a c- by these averagepositions, plus the M-sites which lie VEBLEN AND BURNHAM.' BIOPYRIBOLES 1059

Table 4c. Selectedbond distancesand anslesin chesterite

o o M-0 Distances, A Si-o Distances, A SiT3ATetrahedron MT1 Octahedron MT2 octahedron l{I3 Octahedron SiTIA Tetrahedron SIT2ATetrahedron Sir3A-oT5A I.623 r.ff1-oT2A(6) 2.05 MI2-OHTA 2.10 MI3_OHTA 2.10 sirlA-or1A 1.608 sir2A-or3A 1.639 Sir3A-or5A 7.667 2.01 MT2-OHTB 2.06 lfr3-oHTB 2,01 sirrA-oTlA(1) 7.627 sir2A-or4A L.629 Ifrl-or2A(7) SiT3A-0T74 L.669 Mr1-or2B(8) 2.lI MT2-OT2A 2.77 lfr3-or4A 2.05 SiTIA-OT2A I. 640 S1T2A-OT5A 1;618 51T2A-0T6A 1.609 Sir3A-OT8A 1.551 Ifrl-or2B(9) 2.12 Ifr2-0T28 2.10 MT3-or4B 2.08 SiTIA.OT3A L.591 Lfrt-oHTA 2.O2 MT2-OT4A 2.09 MT3-OT7A 2.06 1. 618 Mean L.624 Mean L.628 }fil-OHTB 2-01 MT2-OT4B 2.04 MT3-OT7B 2.16 Mean S1T3BTetrahedron Mean 2.O7 Mean 2.08 Mean 2.O9 SIT1B Tetrahedron SiT2BTetrahedron siTlB'oT1B 1.660 sir2B-or3B L.632 Slr3B-0T58 1.630 MT5 octahedron MDI 0ctahedron MT4 Octahedron sirlB-orrB(1) 1.601 S1T2B-OT4B 1.585 51T38-0T68 I'667 SiT3B-OT7B 1.572 MI4-OT4A 2.16 MT5_OT7A 2.L5 IrDl-oDlA 2.rl SiTlB-OT2B 1.554 SiT2B-OT5B 1.648 L.635 Sir3B-OT88 1.600 MT4-OT4B 2.16 MT5-OT7B 2.20 MDl-OD1B 2.O3 StTlB-OT3B 1. 615 sir2B-or6B MT4-OT7A 2.Ol MI5-0D4A 2.04 MDt-0D2A 2.16 1.625 L"617 MI4-OT7B 2.IO MT5-OD4B 1.99 MD1-OD2B 2.II Mean 1. 608 Mean MT4-OD4A 2.00 MT5-OD6A 2.42 MDI-OHDA 2.TI MT4-OD4B 2.02 MT5-OD6B 2.82 MDI-OHDB 2.08 S1DIA Tetrahedron SiD2A Tetrahedron SiDIA-OD1A 1.589 siD2A-0D2A 1. 559 Mean 2.lO Mean 2.09 Mean 2.27 SiDIA-OD5A 1.s99 SiD2A-OD4A I.647 S1D1A-OD6A r.646 siD2A-OD5A 1.622 SiDIA-OD7A 1.616 siD2A-oD6A 1,680 MT5-OD5B 2.92 2. siD2A-MT5 91 Mean 1.613 Mean 1.621

MD2 Octahedron MD3 Octahedron MD4 Octahedron SiDIB Tetrahedron SiD2B Tetrahedron MD2-ODlA 2.72 2 MD3-ODlA 2.10 rfD4-oD2A 2,L9 siDlB-oDlB 1. 689 s1D2B-OD2B r.655 1. 618 MD2-ODIB 2.LI 2 MD3-OD1B 2.IO MD4-OD2B 2.13 siDlB-OD5B 1.604 siD2B-OD4B 1.655 MD2-OD2A 2.10 MD3-OHDA 2.04 MD4-OT8A 2.02 siD1B-OD6B r.636 siD2B-oD5B 1.603 MD2-OD2B 2.06 MD3_OHDB2.09 MD4-OT8B 2.O2 SiDIB_OD7B L.607 siD2B-0D68 MD2-0T8A 2.07 MD4-0T6A 2.44 I.634 Mean 7.633 MD2-0T88 2.06 Mean 2,09 MD4-OT6B 2.8L Mean o = Mean 2.O9 Mean 2,21 Estimated Standard Error Si-o 0.018A

MD4-OD5E 2.96 s1T3A-MD4 2. 87 o Estlmated Standard Error M-0 = 0.02A

*Coordinate lransfomations indicated in Darentheses: Tetrahedral Chains (L) x,4-y, \+z (6) x, y, z-1 A Triple Chaln B Trlp1e Chain (2) La+x, \- y, 7-z (7) x, ta- y, z-\ orlA-or1A-or1A 179.60, OTIB-OTI-B-OT1B167.30 (3) x, y, 3/2-z (8) *- 'z-y, 'a- z %- L , 015A-0164-015A 173.50 0T58-0T68-0T58 161,10 (4) -x, -y, 1-z (9) x-% , y, L-z (5) x, -y, ta+z A Double Chain B Double Chaln Atoms with no transfomation indlcated are at aAz, or OD5A-OD6A-OD5A170.60 0D5B-0D6B-OD5B161.20 are participating in an unanbiguous distance or angle. Numerals preceeding bonds lndicate nulEiplicity. Estimated Standard Error 0-0-0 = 1.30

on 2-fold axes, has C2/c symmetry and is a triple- sured from two (100) lamellae in a jimthompsonite chain analog of high clinoenstatite.It contains l7 crystal. Becausethe reciprocal lattice points of the atoms in the asymmetricunit and is topologicallythe jimthompsonite and clinojimthompsoniteoverlapped same as the NaMg-silicate structure of Drits et al. for / : 4n, those diffractionswere omitted. In spite of (1974). Site nomenclaturein this model is analogous long counting times, the measured intensitieswere to that for the jimthompsonite structure,except that very low; thus a large number of diffractions were "A" and "B" are dropped becauseall triple chains unobserved. The absorption correction was calcu- are now symmetricallyequivalent. lated assumingthat the entire crystal was clinojim- thompsonite,but becausethe transmissionfactors for Structure rertnement the entire crystal were high and only varied between Becauseclinojimthompsonite occurs as lamellaein 0.88 and 0.93,intensity errors introduced by errorsin jimthompsonite, a good single crystal could not be the absorption correction are small' found. Consequently,X-ray intensity data were mea. In order to check the space-group extinctions, all 1060 VEBLEN AND BURNHAM: BIOPYRIBOLES

Table 6. Clinojimthompsonite atomic coordinates,temperature refinementin C2/c convergedreadily. The scalefac- factors, and metal (M) site occupancies tor, atomic coordinates, isotropic temperature fac-

AEm I'el Fe}}'19 tors, and unconstrainedMg-Fe occupanciesof the

UI 0.0 0,0289(2) 0.25 0. l4(r3) 0 .032( 15) M-sites were varied simultaneouslyduring the final M2 0.0 0.0849(2) 0.75 0.16 ( 13) 0.04r(16) cyclesof refinement.Only observeddiffractions were y3 0.0 o. 1428(2) 0 ,25 0. 18(13) 0.09s(r6) u4 0.0 0.2013(2) o.1s 0.1s(14) 0.008( 16) included in the normal equations. Values of R for M5 0.0 0.2s69(r) 0.25 0.43(7) 0.845(r8) SiI 0.2909(4) 0,0277(2) 0.7656(6) 0.18(5) clinojimthompsonite are listed in Table 2; atomic si2 0.2880(4) 0. r395(2) o,7610(7) 0.24(6) s13 0.297r(4) 0.1,962(2) 0.2630(7) 0.2s(6) positions, isotropic temperature factors, and occu- OH 0. r153(8) 0.0850(3) 0.1s66(13) 0.23(12) OI 0.3499(11) 0.0037(3) 0.5413(r7) 0,42(r5) panciesare given in Table 6; and selectedbond dis- 02 0. l1s4(9) 0.0278(4) 0.6s52(15) 0.47(r3) tancesand anglesare compiled in Table 4b. Tables5 o3 0.344e(8) 0.0840(3) 0.8197(13) 0.33(11) o4 0.1149(10) 0. r420(3) 0.6555 ( 15) 0. 48(r4) and 8 (seefootnote 2) contain more complete bond 05 0.3s35(i0) 0.1713(3) 0.0344(15) 0.5e(14) 06 0.3523(10) 0.1628(3) 0.5390(r6) 0.76(r5) distancesand anglesand the observedand final calcu- o7 o. 1219(9) 0.1983(3) 0.r536(15) 0.4s(14) 08 0.3771(9) 0.2474(3) 0.3581(3) 0.48(15) lated structure factors for those reflectionswith 1 ) 3ot. classesof diffractions were measured.Of 2253 h + k The clinojimthompsonite structure : 2n * I diffractions, 142 or 6.3 percent were ob- Figures 5 and 6 show projections of clinojim- served (greater than 2o). If these diffractions were thompsonite and presentthe site nomenclature.The truly extinct,about 5 percentwould be expectedto be structure consists of l-beams containing five crys- observableat the 2o level.This is closeto what was tallographically distinct M-sites and symmetrically observed, but refinement was neverthelessfirst at- equivalent,O-rotated triple silicatechains with inner tempted in spacegroup P2'/c, utilizing Iheh -l k:2n and outer rotation anglesof 171.4"(Ol-Ol-Ol) and * 1 diffractions. Atoms that would have been sym- 170.0' (O5-O6-05). TheseI-beams are connectedin metrically related to each other in C2/c began to a ***i stacking sequence,analogous to the stack- move toward C2/ c-equivalentpositions, coordinate ing in monoclinicpyroxenes and amphiboles. correlation coefficients between these atoms became As in the jimthompsonite structure, the inner M- fairly high (about 0.7), and unacceptablebond dis- sitesare quite regular, but slightly compressedin the tancesdeveloped. Calculated structure factors for the [00] direction, while M5, the outer M-site, is a dis- h + k : 2n * | diffractionsbore no obviousrelation- torted octahedron.Fe is again concentratedover Mg ship to the low observedstructure factors, and R, did in this distorted outer site; unconstrainedoccupancy not fall below 10.8percent. It was concludedthat the refinement resultedin Fel(Fe + Mg) : 0.20, while structure must have C2/c symmetry. (Fe + Mn)/(Fe * Mn * Mg + Ca * Alo"r) :0.22 When the criterion for consideringa structure fac- was measuredwith the electronmicroprobe from the tor to be observed was increasedto 3o (of the in- clino lamellae.The I-beams of clinojimthompsonite tegrated intensity), only 12 h + k : 2n l- I diffrac- show the same type of chain warping as those of its tions remained. These were removed, and the orthorhombicpolymorph.

Fig. 5. Projectionof clinojimthompsoniteonto (100).The triplechain is O-rotated VEBLEN AND BURNHAM: BIOPYRIBOLES t06l

o

Fig.6. c-axisprojectionoftheclinojimthompsonitestructure.Thestackingsequenceis****,andalltriplechainsaresymmetrically plane. equivalent.As in jimthompsonite, the basal oxygen layersof the silicatechains are warped out of the (100)

Chesterite There are 5l atomsin the asymmetricunit of this structure. Model deriuation Atom nomenclaturein the triple-chainI-beam is An l/(z) testusing all measuredchesterite structure the sameas that for jimthompsonite,and the nomen- factors resultedin a curve that is acentric at low claturefor the double-chainI-beam follows the stan- valuesof z and hypercentricat z valuesabove 0.3 dard orthoamphiboleterminology (Finger, 1970). (Figure 7). Becausemcire than half the data were Atoms in the triple-chainI-beams are denotedby unobservedand had statisticallyassigned Fi6" values "T," and thosein the double-chainl-beams by "D." correspondingto I^1n/3, the intensityaverages used Theseletters are insertedafter the chemicalpart of in this test are probablyincorrect, and the test in- theatom names; for example,OT5A is in theA triple valid.In addition,the proposedmodel for chesterite chain,while OD5A is in the A doublechain. containeda pseudo-inversioncenter. The testresults Structure refinement were thereforeignored, and a detailedmodel was constructedin the acentricspace group A2rma,the After refining only the scalefactor for one cycle, only spacegroup that is consistentwith both the R. for all observedreflections was 0.074,indicating diffraction symbol and pyribole I-beam symmetry that the modelcalculated in the abovefashion was and linkageoperations. extremelyclose to the real structure.After refining The chesteritemodel contains double- and triple- scalefactor and atom positionsfor two cycles,con- chain I-beamsalternating in the b directionand vergencewas achieved,and R, reached0.060 after stackedwith a ++-- sequencesimilar to that in only minor adjustmentsin atomicpositions. other orthopyriboles.Coordinates for an initial struc- Isotropic temperaturefactors and M-site Fe-Mg ture model were calculatedfrom the refined atom coordinatesof jimthompsonite.The triple-chainl- beam of chesteritewas taken intact from the jim- thompsonitestructure, and the doublechain was con- structedfrom the Si2, Si3, M2, M3, M4, and M5 polyhedraof the adjacenttriple-chain I-beam in jimthompsonite.The following transformation of the appropriate jimthompsonite atoms yielded the model:

.xg6 : I1t y"5:0.6yp*0.1 Fig. 7. Plot of the chesteriteN(z) test. All measured reflections consistent with diffraction symbol mmmA"a were utilized. See Z.y,: Zy + 0.25 text for interDretation. VEBLEN AND BURNHAM: BIOPYRIBOLES occupancies were then refined, and nu fell to 0.059 atoms in the A triple chain still had negativeisotropic for all observedreflections. Atoms in the triple-chain temperature factors close to zero. These temperature I-beam were, however,strongly correlatedwith cor- factors were significantly correlated to the temper- responding atoms in the double chains, especially ature factors of corresponding atoms of all other with respect to thermal parameters, which had corre- tetrahedra. It was decided, however, not to pursue lation coefficientsas high as 0.8. Analogous atoms any further constraint schemes,because full-matrix within the following polyhedra were strongly corre- least-squaresrefinement of a 5l-atom structure re- lated: SiDIA and SiT2A; SiD2A and SiT3A; SiDIB quireslarge amounts of computer time, and the abso- and SiT2B; SiD2B and SiT3B; MD3 and MT2; MDI lute values of these temperature factors were much and MT3; MD2 and MT4; and MD4 and MT5. For lessthan their standarderrors. The changesin atomic example, ODIA, which is the apical oxygen of the positionswhich resultedfrom the thermal constraints SiDIA tetrahedron, was strongly correlated with were insignificant(less than 0.lo). OT4A, the apical oxygen of the SiT2A tetrahedron. Final R values for chesteriteare given in Table 2, In addition, atoms of the SiTIA and SiTIB tetra- and atomic coordinates,isotropic temperature fac- hedra exhibited strong correlationswith those of all tors, and M-site occupanciesare listed in Table 7. the double-chaintetrahedra. and MTI was correlated Becausethe positional parameterswere nearly inde- with all other regular M-sites in the structure. Be- pendent of the thermal and occupancy parameters, causethe abovepairs ofpolyhedra havenearly identi- the positional standard errors should be realistic.A cal shapesand environments with respectto first- and few of the refinedbond distances,such as the SiT3A- second-nearestpolyhedra, a pseudo-inversioncenter OT8A distance of 1.55(2), suggest,however, that existsat (l/2,2/5, l/4), and there is a pseudo-2,axis these errors are larger. Becausethe thermal con- in the B layers and a pseudo-glideplane in the A straints affectedoccupancies by as much as 3oo",the layers; this strong pseudo-symmetryis probably re- occupancystandard errors are probably severaltimes sponsiblefor the high correlations.Similar behavior too low. Likewise, the isotropic temperature factor is often noted when refining a structurethat deviates errors are undoubtedlytoo low, being basedon plau- only slightly from a higher space-groupsymmetry sible but unproven constraints that eliminated the [see,for example,Bailey (1975)and Guggenheimand largest covariances.The temperature-factor errors Bailey (1975)1. In chesteritethe triple and double were approximately halved by applying the equality chains are topologically distinct, and the pseudo- constraints. symmetry is thereforeonly local. Becauseonly a small chesteritecrystal suitablefor The isotropic temperaturefactors of some of the intensitymeasurement could be found, the number of correlated atoms behaved erratically during the re- unobservedintensities was large; only 2633 out of the finement,one of a pair assuminga high positivevalue t 6043 measured intensitieswere observed.Structure and the other becomingnegative. Because the corre- factor calculations for the unobserved diffractions lated sites were so similar with respect to both first- produced no largePs inconsistentwith the structure. and second-nearestneighbors, it was assumedthat The structurerefinement of chesteritecould undoubt- the atoms occupying them would have similar ther- edly be improved if a monophasesingle crystal large mal parameters,and the isotropic temperaturefac- enough to yield a high ratio of observed to unob- tors of correspondingSi, O, and OH ions in the triple serveddiffractions could be found. and double chains were constrainedto be equal. M- Selectedchesterite bond distancesand anglesare site atoms were left unconstrained becauseof the presentedin Table 4c. Table 5 contains more com- additional correlationsbetween occupancy and tem- plete distancesand angles,and the observedand cal- perature factors.The temperaturefactors of the con- culated structure factors that were used in the refine- strainedatoms attained more reasonablevalues dur- ment are listed in Table 8 (seefootnote 2). ing this stageof refinement,but M-site temperature factors were still unreasonable.Corresponding M The chesterite structure atoms in the D and T l-beams,as listed above, were Chesteriteis the first known chain silicatehaving therefore also constrained to have equal isotropic more than one set of topologically-distinctchains. temperaturefactors. This resultedin more reasonable The structure is shown in Figures8, 9, and 10. There thermal parametersand little changein Mg-Fe occu- are two distinct typesof I-beams,one similar to the I- pancy; the largest change in an occupancy was 7.6 beam of anthophyllite, with A and B double chains, percent, about 3oo".The two unconstrainedoxygen and the other very close to the I-beam of jim- VEBLEN AND BURNHAM.' BIOPYRIBOLES 1063

factors, and thompsonite, with A and B triple chains. All the Table 7. Chesterite atomic coordinates, temperature metal (M) site occuPancies chains are O-rotated, except that the inner rotation angle of the A triple chain differs from l80o by less I'e / I'efMg jimthompsonite and than its standard error. As in Mrr 0.12610* 0.2675(2) 0.1451(13) 0.38(10) 0.090(15) clinojimthompsonite, the outer tetrahedra of both r'f!2 0,1224(7) 0.3013(2) 0.6413(15) 0.82(7) o. r89(rs) MT3 0.t264(6) 0.3357(1) 0.i4s8(14) 0.30(6) 0.084(12) triple chains are more rotated than the inner ones. !fr4 0.1243(6) 0.3715(2) 0.6423Gs) 0.33(7) 0.028(13) l'n5 o.1233(5) 0.4044(t) 0.1414(7) o.4r(3) 0.805(13) The stackingsequence of the I-beamsis **--, the sirrA0.2704(6) O.2668(2) 0.8090(1r) 0.36(9) sir2A 0.269r(6) 0.3344(2) 0-8r69(12) 0.44(5) same as for orthopyroxenes,orthoamphiboles, and sir3A c.2732(6) O.3682(2) 0.3r68(12) 0.43(5) jimthompsonite. is acentric, sirrBo.481l(6) O.2670(2) 0.s184(12) O.49(9) Becausethe structure sir2Bo.4815(6) 0.3334(2) 0.5245(11) 0.35(5) there are two nonequivalent ++-- stacking ar- s11380.4755(6) 0.3678(2) 0.0219(r2) 0.43(5) onrA 0.183s(r0) 0.3oos(4) 0.3086(34) 0.59(14) rangements that are probably equally likely; the ollTB 0.5695(10) 0.30I5(4) 0.0181(34) 0.44(13) 0r1A 0.2990(8) 0.2499(3) 0.56r6(24) -0.03(17) atomic positionsin Table 7 are for one of these.As in or2A 0.1823(8) O.2672(4) o.8r i2 (26) -0.06 ( 20) or3A 0.29i6(9) 0.3002(4) 0.800r(30) 0,74(r3) other low-Ca pyriboles, is concentrated over or4A 0. 1817(9) 0.3350(4) 0.8120(28) 0.59(14) in the distorted outer M-sites, and all 0'r5L o.2992(9) 0.3490(4) 0.0750(27) 0.e2(13) magnesium 0T5A 0. 3023(9) o. 3521(4) 0.s806(28) 0.91(13) four independentchains are warped out of the (100) orTA 0.1836(9) 0.3694(3) o.3r30(30) 0.64(r3) 0r8A 0.3133(8) 0.3982(4) 0.3317(30) 0.37(rl) plane. Occupancy refinementgave overall Fe/(Fe + 0rlB o. 4s23(10) o .2468(4) 0.2771(29) 0,97(25) 0r2B o. 564s(9) O.2664(4) 0. s191(26) 0.28(20) Mg) : 0.24 for the structure,while electron micro- dr3B 0.4s03(9) 0.3001(3) 0.484r(27) 0.39(11) tJr4B 0.s667(9) 0.3349(4) 0.5234(28) 0.51(13) probe analysisgave (Fe + Mn)/(Fe * Mn * Mg + orsB 0.4497(9) 0.3551(3) 0.30r0(26) 0.43(10) : 0.26. 0168 0.450s(9) 0.3454(3) 0.794s(28) 0.59(11) Ca) orTB 0.5599(9) 0.3688(3) 0.0326(29) 0.62(13) 0T8B 0.4345(9) 0.3980(4) 0.964s(30) 0.68(12) A-site occupancies luD3 0. 3718(8) 0. s 0. 1485(22) C. 82** 0. 214 ( I 9) Fourier synthesesand ]lDl 0.3771(5) 0.46s4(2) 0.6443(16) o.30 o.os7(13) r,rD2 0.3749(5) 0.4296(2) 0. r489(18) 0-33 -0.017(13) To confirm the refinementresults, electron density yD4 0.3752(s) 0.3963(1) 0.6457(9) o.4l 0.713(13) siDlA0.2302(6) o.4663(2) 0.3203(13) 0.44 difference maps were calculated for all three refined siD2^ 0.2282(6) O.4323(2) 0.8125(14) 0.43 Preliminary difference maps of jim- siDIB 0.s201(6) o.t 664(2) 0.9796(r3) 0.35 structures. siD2B0.5244(6) o.4336(2) 0.4787(r4) 0.43 thompsonite and clinojimthompsonite, calculated 0HDA0.3187(13) 0.5 o.8r17(46) 0.59 olrDB0.4337(13) 0.5 0,4778(47) 0.44 prior to the applicationof anomalousdispersion cor- oDTA 0.2017(r3) 0. s o.3064(44) O.74 oDlA 0.31s4(r0) 0.4647(4) 0.3101(32) 0.69 rections,showed erratic differencesas large as 3.5 e/ oDsA o.2009(10) 0.4522(4) 0.5770(30) O.92 oD6A 0.1974(9) 0.4476(4) 0.0804(30) 0.91 A3; these differences were eliminated by further re- oD2A 0.3r19(9) 0.4316(4) 0.8201(34) 0.64 finement using scattering factors corrected for anom- oD4A 0.1858(8) 0.4004(4) 0.828r(26) 0.37 }DTA 0.5442(r2) o.5 0.0268(40) o.39 alous dispersion,even though R values were not im- oDrB 0.4294(9) 0.4653(4) 0.9804(31) o.sl oDsB 0.ss13(9) o.4457(4) 0.1995(31) 0.43 proved. oD6B 0.5480(9) 0.4s53(4) 0.7028(32) 0.59 oD2B 0.4357(io) 0.4309(4) 0.4738(33) 0.62 The final jimthompsonite differencemap showsno ooas o.soba(s) o.4o3s(4) 0.5367(26) 0.68 uninterpretablediscrepancies larger than 1 e,/A3.Pos- * Constrained to fix otigin. itive and negativedifferences indicative of anisotropic ** Temperature facEors wiEhout erlors were constrained to equal t.*p"."tr." factors of corresponding atons in the triPle_chain thermal motion are associatedwith M5 (1.8 e/A3) T-bean, and with O5A and 06A (l.l e/A'). These oxygen difference densities are consistent with tetrahedral torsion about the a* direction, a type of thermal motion that is observedin other chain silicates.The amounts of other speciesmust also be presentin the most pronounced differencepeak is located in the A-site, or the analysisfor Na must be too low. position correspondingto the amphiboleA-site. This The final differencemap for clinojimthompsonite is peak is shapedlike a lopsideddumbbell with maxima extremelyflat, with no differencesgreater than *0.6 of 3.4e/ At at (0.37,0.323, 0.89) and I .3e/ A3 at (0.37, e/A3.The A-site appearsto be empty. 0.347,0.89). The peak, contoured in Figure I l, prob- The largest differences exhibited by the chesterite ably signifies positionally disordered partial occu- difference map are again associatedwith the A-sites. pancy of the A-site. The identity of this site's in- The AT site, which is sandwichedbetween the triple- habitants,however, is not so clear;the amount of Na chain l-beams, is slightly split, with two maxima of indicatedby electronprobe analysesofjimthompson- 1.4 e/As focated at (0.36, 0.297, 0.17) and (0'38, ite, including the crystal used for intensity data col- 0.307,0.16).The AD site,on theother hand, has only lection, representsabout 2.5 percent Na occupancy, a singlemaximum of 1.4e/At aL(0.12,0'5' 0.61). and that amount is not likely to produce a difference Careful examinationof the differencemaps did not density peak as large as the observed one. Small locate H+ ions in any of the structures. 1064 VEBLEN AND BURNHAM: BIOPYRIBOLES

Fig. 8. (100) projectionof the chesteritestructure, showing the A-chains. Double (D) chainsand triple (T) chainsalternate in the D direction, and all chainsare slightly O-rotated.

Bond distancesand angles sonite,the two shortestSi-O distancesofthese outer tetrahedra are associatedwith the longest O-O dis- Bond distancesand anglesin the threerefined pyri- tanceand the widest O-Si-O angle;this is alsotrue of boles from Chester (Tables 4 and 5) are strikingly the tetrahedra in many other silicates(Brown and similar to the correspondingdistances and anglesof Gibbs, 1970). low-calcium pyroxenesand amphiboles.In this sec- The outermost tetrahedron of the A-chain in all tion the tetrahedraand octahedrain the structuresof orthopyriboles shares an edge with the distorted the new minerals are compared with those of an- outer octahedral site. The O-O distancealong this thophyllite (Finger, 1970) and enstatite (Morimoto shared edge is 2.50A in enstatiteand anthophyllite, and Koto, 1969). 2.52 in jimthompsonite,and 2.48 and 2.53 in the The means of all Si-O distances are 1.62A for chesteritetriple chain and double chain, respectively. jimthompsonite, 1.63 for clinojimthompsonite, 1.62 The analogousedge in clinojimthompsoniteis 2.54A. for the triple chains in chesterite,and 1.63 for the If the outermost M-sites of the octahedralstrips are chesteritedouble chains.These values are typical for consideredto be six-coordinated,then the outermost silicates,and compare with 1.64,{ for enstatiteand tetrahedronof the enstatiteB-chain sharesno edges, 1.63,{ for anthophyllite. Si-O distancesin the new while those of the jimthompsonite and chesteriteB- structuresappear to be reasonable,except for some chains share one edge.The anthophyllite refined by of those in chesterite,which are as low as 1.55(2)A Finger (1970)is an intermediatecase, because the 6th and may result from correlation problems in the re- and 7th bonds from M4 to oxygen (O5B and 068) finement. The inner tetrahedra (Sil and Si2) in the are the samelength within error. This sharedtetrahe- triple chains of all three structuresare quite regular, dral edgeshortens as it is increasinglyshared by the as are the Tl tetrahedra in anthophyllite, while the outer M-site: the edge length is 2.594 in enstatite, outermost tetrahedra are more distorted, similar to 2.57 in anthophyllite, 2.55 in jimthompsonite, and the T2's in anthophyllite and the tetrahedraof ortho- 2.56 and 2.53 in the chesteritetriple and double pyroxenes. In jimthompsonite and clinojimthomp- chains.

Fig. 9. ChesteriteB-chains in (100) projection Double and triple chainsalternate in the b direction, and all chainsare O-rotated VEBLEN AND BURNHAM: BIOPYRIBOLES 1065

Fig. 10. (001) projection of the chesteritestructure The stackingsequence is f *--, and I-beamswith double and triple chains alternatein the b direction.Double and triple B-chainsare sandwichedbetween octahedral layers of like skew,while doubleand triple A- chainsare betweenIayers of oppositeskew. Basaloxygen layersof all silicatechains are warped out oi the ( 100)plane

As in enstatiteand anthophyllite, all but the out- + + + + stackingpolymorph of chesterite'If this hy- ermost polyhedra of an octahedral strip are quite pothesisis correct,the structurewill havetriple-chain regular in the refined structures.The most notable I-beamsbisected by c-glideplanes parallel to (010)' departure from regularity is shorteningof the edges and double-chainI-beams bisected by mirrorsparal- shared between octahedra. The mean unsharedoc- lel to (010).Such a structurewill havespace group tahedral O-O distancesin jimthompsonite and clino- jimthompsonite are 3.07A, and the mean for chester- c-glide ite is 3.08A; the mean shared edge for all three mineralsis only 2.81A. By comparison,the unshared edges of the Ml octahedron in enstatite average 3.034, and the mean of the sharededges is 2.82A. ln talc, the means of unshared and shared edgesare 3.064 and 2.804 (Rayner and Brown, 1973).The shorteningof sharedoctahedral edges is thus charac- teristic of low-Ca biopyriboles in all structural groups. The M-O distancesin the outer, distorted octa- hedra of the three new structuresare consistentwith occupancy primarily by Fe and Mg. The minimum M-O distancesof thesesites, and of M2 in enstatite and M4 in anthophyllite, are closeto 2.00A (+0.04), much shorter than the analogousdistances of calcic pyriboles;the shortest M2-O distancein diopside is 2.354 (Cameron et al., 1973),and the minimum M4- O distancein tremolite is2.32A (Suenoet al,,1973).

Unnamed mineral Fig. I I Difference electron density in the slightly occupied jimthompsonite A-site. The contours, from outer to inner, The unit-cell dimensions,space group A2/m (or represent 0, l, 2, and 3 e/A3. The form of the density differences A2, Am), and (ft0l) intensitiesof the unnamed min- suggeststhat there is considerablepositional disorder among A- eral strongly suggest that it will prove to be the site occupants. 1066 VEBLEN AND BURNHAM: BIOPYRIBOLES

Am if there are two symmetrically distinct silicate suggeststhat other mixed-chainsilicates intermediate chains in each l-beam, and spacegroup A2/m if the betweenthe pyroxenesand the amphiboleswill be two chains of each l-beam are related to each other found.The most likely phaseswould be PMP'swith by 2-fold rotation. Since Am and A2/m exhibit the spacegroups A2rma or A2/m or theirsubgroups, and samediffraction symmetry,either modification of the havingb = 27A. The possibilityof mineralswith basic structure may be correct. chainswider than triple is alsointriguing; the fibrous talc of Chesterhas not yet beenexamined carefully, The new minerals as biopyriboles and it may be that part of what appearsto be talc is Thompson (1970)explained that most amphiboles really one or more mineralsintermediate between can be thought of as l: I mixtures of a pyroxeneand a jimthompsoniteand talc. mica, sectoredalong the pyroxene c-glidesand the It is likelythat mineralsisostructural with the new mica c-glides into (010) slabs, and reassembled.To biopyribolesfrom Chester,but havingdifferent com- emphasizethe close relationship betweenthese min- positions,will also be found. This possibilityis en- eral groups, he unearthedJohannsen's (l9ll) super- hancedby thesynthesis of aC2/c, NaMg triple-chain group name, "biopyribole." In addition to l:l mix- silicateby Drits et al. (1974,1976).This fibrous phase tures (MP), other slab mixtures,such as MMp, might is reportedto containNa in the M5 siteand OH'- be expected.AII mixtures of this sort are biopyri- replacingtetrahedral O'z-; it is thereforenot a true boles,and all belongto a "polysomatic" seriesof slab biopyribole, but would be similar to "hydro- mixtures residing in pyroxene-mica compositional amphiboles"and could be thoughtof as a slabmix- space(Thompson, 1978).When viewed in this way, ture of talc and NaMg "hydropyroxene."Similar jimthompsonite and clinojimthompsonite are phasesthat are true biopyribolescould exhibit the MMP's, and chesterite(and probably the unnamed wide rangeof ionic substitutionsobserved in the py- mineralas well) is an MMPMP. roxenes,amphiboles, and micas,as could minerals For an infinite crystal, there are infinite possible isostructuralwith chesterite.The new mineralsfrom polysomes, but the known pyriboles all fall in the Chestermay simplybe low-calciummembers of two followinggroups: (l) thosewith pure odd chainsof large,but heretoforeunrecognized, mineral groups. one type (all single chains, or all triple chains, for It is expectedthat the orthorhombicstructure types example);(2) pure even chains(all double, all quad- will be restrictedto compositionscontaining only ruple chains, etc.); or (3) mixed alternating odd and small amountsof Ca or other largeoctahedral cat- even chains for which the chain numbers differ by I lons. (mixed double and triple chains, or mixed singleand doublechains). Ideal "pure odds" will all havepoly- Polytypism some formulas with one P and an even number of Pyribolepolytypism has long been a subjectof M's, and will have spacegroups pbca or C2/c, or interestto mineralogistsand petrologists.The com- their subgroups, for the two common polytypes. positionaland intensive-parameter stability ranges of "Pure evens" will have formulas with one p and an the enstatitepolymorphs, and of anthophylliteand odd number of M's, and will havemaximum symme- ,are still not agreedupon. Although tries of Pnma or 12/m. Mixed odd-evens will have they in no way solvethe stabilityproblem, the new two P's and an odd number of M's in the oolvsome pyribolesfrom Chesterare of interestin that they formula, and will havemaximum symmetri"r2Zr*o possessthe two mostcommon stacking sequences of or A2/m. Approximate 6-axis lengthsof thesepossi- the pyroxenesand amphiboles:chesterite and jim- bilities can be found by multiplying the number of thompsoniteare stackedin the samemanner as the symbols in the polysomeformula by 9,A,.These three orthopyroxenesand orthoamphiboles,while clino- possibilitiesalso appearto be the most likely because jimthompsoniteand probably the unnamedphase the mineral reaction sequenceat Chesteris pure even have stackingsequences analogous to thoseof the - mixed even-odd + pure odd; no possibilitiessuch clinopyroxenesand clinoamphiboles. as MMPMMPMP have been observed,and the se- The new orthorhombicpyriboles have **-- quencealso suggeststhat phasessuch as the MMp's stacking,while the new clinopyribolesare stacked are preferred over MPMMMP's, which would have + + + + (seeVeblen et al., 1977, for an explanationof the same composition. this stackingterminology). No pyribolesfrom Ches- The existenceof chesteriteand the unnamed min- ter were found to have a proto stackingsequence - - eral, consisting of mixed double and triple chains, (+ + ). In addition,none of the 88 singlecrystals VEBLEN AND BURNHAM: BIOPYRIBOLES t067 examined by precessionphotography showed evi- A-chains to have alternating O- and ^S-rotations denceof streakingin thedirection of a*, whichwould (Thompson, 1970). These alternative space groups signify fine-scalestacking disorder. would, however,be readily differentiatedwith single- crystal X-ray methods.An orthopyroxenewith P2'ca parity Pyribolechain rotations and the rule symmetry has been reported (Smyth, 1974),but the Tetrahedralrotations play an important role in intensities of equivalent violating reflections (the sheetsilicates (Bailey, 1966; Bailey et al., 1967).In {071} diffractions, for example) are not equal in the micas,the primary influenceon the degreeof rotation published precessionphotograph and do not even is apparentlythe requirementthat the tetrahedraland possessinversion symmetry. Furthermore, ortho- octahedrallayers occupy the samearea (McCauley pyribole refinementsof averagestructures should ex- and Newnham.l97l). In additionto thesedimen- hibit strongly anisotropic thermal motion, which is sional requirements,the tetrahedralrotations of not observed.Chesterite has been refined in a space chain silicatesare further subjectto restrictionsaris- group that does not require the parity-violating A- ing from connectionsbetween I-beams. One formula- chainsto havethe samesense or degreeof rotation in tion of theseadditional requirements is theparity rule adjacent chains, yet the parity rule is still violated (Thompson,1970; Papike and Ross, 1970),which with O-rotated A-chains having chain anglesof 173.5' statesthat in idealpyriboles having regular polyhedra and 170.6'. We thereforeconclude that orthopyroxene exceptfor the outer M-site, geometricalconstraints and jimthompsonite refinementsin Pbca, and ortho- are imposedon tetrahedralrotations by the octahe- amphibole refinementsin Pnma, are correct. dral stacking sequence.Chains betweenoctahedral layersof like skew(here defined as '-chains)must be The cause of orthopyribole A'chain rotations straight,or havethe same rotation sense, while chains betweenlayers of oppositeskew (defined as x-chains) The triple chains of the new mineralsfrom Chester must be straight,or haveO- andS-rotations alternat- provide new information on the causesof pyribole ing in the6 direction.This ruleis rigorousfor regular chain rotations. In all casesthe interior and more polyhedraand must be strictly obeyed, even for small mica-like part of the chain is less rotated than the tetrahedralrotations. The polyhedra ofreal pyriboles outer part, suggestingthat the rotations, at least of arenot perflectlyregular, however, and in thissection the outer tetrahedra, result primarily from the re- we will showthat violationsof the parity rule are a quirements of inter-I-beam connection geometry. necessaryconsequence of distortions arising from The same effect is observable in the amphiboles, edge-sharingamong polyhedra. though not so graphically as in the pyriboles with triple chains.In the following discussionwe therefore Rotationsand symmetryof real orthopyriboles concentrate on the outermost polyhedra along the It hasbeen pointed out that theA-chains ofrefined edgesof pyribole I-beams. orthopyribolesand the chains of protopyriboles, If the chains betweenoctahedral layers of opposite which arelocated between octahedral layers of oppo- skew (x-chains)are all equivalent,as they are in real site skew,are all O-rotated,and that they therefore pyriboles,and if the inner M polyhedra are perfectly violatethe parity rule. Papikeand Ross(1970) ex- regular, then the two non-bridging oxygens of the plainedthat the parity violationsof gedritesare ac- outermost tetrahedron of the chain must project or- companiedby M2 distortion,and that the A-chains thogonally onto the same point of any line that de- areless rotated than the B-chains. Papike et al. (1973) scribesthe chain direction (the c-axis, for example). discussedparity violationsin orthopyroxenesand The appropriate atoms are OIA and O2A for the concludedthat theviolations are minimized by exten- pyroxenes,O2A and O4A for the amphiboles, and tional rotation (straightening)of the chains.They O7A and O8A for pyriboleswith triple chains.In the alsosuggested that the outer-M-site(M2) coordina- orthorhombic pyriboles, this requirement simply tion requirementsnecessitate the A-chain O-rota- means that these atom pairs must have the same z tions.Sueno et al. (1976)further explain that poly- coordinate. In a real orthorhombic pyribole, the hedraldistortions permit parity violations. amount of "misfit" exhibited by the outer x-chain It hasbeen suggested that parity violationsdo not, tetrahedron can be defined as the difference. in A. in fact,occur and that the spacegroups ofreal ortho- betweenthe z coordinatesof thesetwo atoms. This pyroxenesand orthoamphibolesmay be P2tca and misfit must be absorbedby distortion of the regular P2rma,rather than Pbca andPnma, thus allowing the inner M octahedra. 1068 VEBLEN AND BURNHAM: BIOPYRIBOLES

fectlyregular. The only exceptionsto this patternare the gedrites. Figure l3 comparesthe A-chain of XYZ ortho- pyroxene(Burnham et al., l97l) with a chainof ideal tetrahedrahaving the samerotation angle. [t is clear that distortionof the pyriboleouter tetrahedraac- countsfor the nonidealmagnitude and sign of the A- chain misfit,and it is apparentthat this distortion arisesto a large degreefrom edge-sharingwith the outer M-site.The O-O distancefor this sharededge is closeto 2.50Ain all refinedorthopyriboles; this is a typical value for sharedpolyhedral edges in many

B-ChoinRototion (O-O-O) typesof compounds(Pauling, 1948, p. 400). It is r?o. t?o. tFo. tgo. r4o. certainlynot surprisingthat the outer tetrahedraof Fig. 12. z-coordinate difference (d) between the two non- orthopyriboleA-chains are distortedby edge-shar- bridging oxygens of the outermost A-chain tetrahedron in refined ing,and if distortionsof the innerM octahedraare to proto- and orthopyriboles, plotted against the A-chain rotation be minimized,then it is essentialthat theA-chains be angle. The difference, d, is a "misfit" parameter for the A-chains O-rotated.This is indeedthe case. and must be absorbed by distortion of the regular octahedra. The line marked "ldeal Tetrahedron" shows the amount of misfit that The only refinedpyribole structures with tetrahe- would be produced by rotation of a chain of regular tetrahedra; dral misfithaving the same direction as that predicted most orthopyriboles have misfits with opposite sign from that by an ideal polyhedralmodel are thoseof gedrite predicted. The points marked "g" refer to gedrite structures (Papikeand Ross,1970). These are the only phases (Papike and Ross, 1970),and the points connectedby continuous and broken line segments refer to orthopyroxene structures at severaftemperatures (Smyth,1973; Sueno et al.,1976: respectively), the chain becoming straighter with increasing temperature. Other data are from Burnham (1967), Bwnham et al. (1971), Finger (1970), Ghose (1965), Gibbs (1969), Kosoi er a/. (1974),Morimoto and Koto (1969),Morimoto eral. (1975),Smyth (197f), Smyth and lro (1977),Takeda (1972), and this paper.

The amount of misfit producedby a regular tet- rahedronis 6:2\/2/3 (T-o)sin.l(t80"- RA) where(T-O) is the tetrahedralcation-oxygen dis- tance,and RA is the rotationangle as traditionally measuredfor chain silicates(e.g. O3-O3-O3 for py- roxenes).The expectedmisfit for T-O : 1.624 is plottedas a functionof rotationangle in Figure12, alongwith the realmisfit exhibited by the x-chainsof orthorhombicpyriboles with dominantly divalent cationsin the outerM-sites. No attemptwas made to includestandard errors in the diagram,but theerrors aresmall enough in all casesso that the conclusions ldcol Tctrohod?c arenot XYZ Orthopyronn€ altered. A-choin Rototion. 169" Figure l2 showsthat the tetrahedralmisfit in real Fig. 13. A.real orthopyroxeneA-chain with rotation angleO3- pyribolesgenerally has the oppositesign and is much O3-O3 of I69'(Burnham et al., l97l\, comparedwith a chain of smallerthan that predictedfor chainsof regulartet- ideal tetrahedra having the same senseand degreeof rotation. The rahedra.This meansthat the misfit would be in- short shared edges of the real tetrahedra are marked "S." The creasedby extendingthe chains, and decreased figure shows how tetrahedral distortion accounts for the small by misfit (6) in the real A-chain, so that straightening of the chain rotating them further.Consequently greater, not less, would necessitateincreased. rather than decreased.distortion of O-rotationwould permit the inner M-sitesto be per- the regular M octahedra. VEBLEN AND BURNHAM: BIOPYRIBOLES r069 plotted in Figure 12 with significantA-site occu- pancy,and the differencein the tetrahedralmisfit of the gedriteA-chains is consistentwith the ditrigonal distortionthat is expectedto arisefrom A-sitebond- ing requirements. o3 6(l) B-chain rotations in orthopyriboles o2 The parity rule statesthat for tetrahedralchains betweenoctahedral layers of like skew(.-chains), all chainsin a layermust have the same sense and degree of rotation. In all known orthopyribolesexcept chesteriteall of the '-chains(B-chains) are symmetri- callyequivalent, and, except for this equivalenceand rqo. r1o" r?o' rqo' the requirementthat the chainsand octahedral strips Fig. 14. z-coordinate difference (d) between the two non- of an l-beam must fit together,there are no rigid bridging oxygens of the outermost B-chain tetrahedron in refined geometricalconstraints on rotations.Inter-I-beam orthopyriboles,plotted againstthe B-chain rotation' Most of the misfit constraintsdo not apply to orthopyriboleB- structures have 6 slightly larger than an ideal tetrahedron with the chains;hence they can assume larger rotation angles same rotation angle, owing to tetrahedral distortion. The two "g" gedrites, and points connected by thanare permittedin the A-chains. points marked are continuous and broken line segments refer to orthopyroxene Figure 14 showsthat differencesin outer non- structures at several temperatures (Smyth, 1973; Sueno et al.' bridgingoxygen z coordinatesin orthopyriboleB- 1976), the chain becoming straighter with increasing temperature. chainsare much betterpredicted by chain rotation Other data are from the same referencesas those of Fig. 12' anglethan they are for theA-chains, as a resultof the greaterregularity of the B-chainouter tetrahedra. Outer M-site coordination There appearsto be a tendencyfor the differenceto If the outerM-sites of pyribolesare considered to beslightly larger than predicted, resulting partly from be six-coordinated,there are two distinctcoordina- shorteningof a tetrahedraledge. It shouldbe noted tion typesfor thesesites. This is perhapsbest demon- that thisedge is sharedwith the outerM-site in some strated by a series of high-temperatureortho- orthopyriboles,but not in others,as discussed in the pyroxenerefinements by Smyth(1973), who showed nextsection. The edgeshortening suggests that there that thereis an M2 coordinationswitch from O3Bto is somedegree of bondingbetween the outer M-site 03B' between700o and 850oC.This behaviorwas andthe distantbridging oxygen of the outertetrahe- confirmedin orthoferrosiliteby Suenoet al. (1976). dron, evenwhen this oxygenis not oneof the metal The higher-temperaturecoordination involves the ion'ssix or sevennearest neishbors. sharingof not only an A-chaintetrahedral edge, but alsoa B-chainedge, and is a configurationsimilar to Conclusionson pyribole chain rotations that in protopyroxenesand C2/c clinopyroxenes.The Becausethe realmisfit in pyribolex-chains is small, coordinationchange is analogousto theone observed and sincethe rotation of thesechains reduces the to take place during the P2'/c-C2/c clinopyroxene misfit,it can be concludedthat theserotations arise transformation(Smyth and Burnham,19721, Brown largelyas a result of tetrahedraledge-sharing dis- et al., 1972). tortions.The smaller rotations of theinner tetrahedra The gedritestructures refined by Papikeand Ross of triple chainssuggest that intra-I-beameffects tend (1970)have outer M-site coordinations that areclose to straightenthe chains, preventingthem from to thoseof low-temperatureorthopyroxenes. In con- achievinga zero-misfitconfiguration. If it is recog- trast, the anthophyllitestructure refined by Finger nizedthat this misfit is generallysmall in the pyri- (1970)exhibits an intermediateconfiguration in that boles,then it mustalso be realizedthat parityviola- O5B and 068 areequidistant from M4. tionspose no problemfor thereal structures. Because The outer M-sitesof jimthompsonite(M5) and thereare no majormisfit difficulties, there is no com- chesterite(MT5 and MD4) possessthe high-temper- pellingreason to expectthe existenceof ortho- and ature orthopyroxeneM2 configuration.They share protopyriboleswith alternatingO-rotated and S-ro- tetrahedraledges with both the A- and B-chains.On tatedx-chains. purelygeometrical grounds it is obviousthat those 1070 VEBLEN AND BURNHAM: BIOPYRIBOLES orthopyriboles with straighter B-chains will tend to Chain warping the higher-temperatureconfiguration, while those Basaloxygen layersof all the chainsin the Chester that have more kinked chains will tend to the lower- biopyribolesare "warped" out of the (100) plane, as temperature outer M-site coordination. All ortho- can be seenin Figures 4, 6, and 10. The same phe- pyriboles refined at room temperaturewith B-chains nomenon also occurs in pyroxenesand amphiboles. straighterthan thoseof Finger's(1970) anthophyllite A measure of the degree of warping in ortho- : (058-068-058 157.5') do havethe higher-tem- pyriboles is the maximum difference,in A, between peratureouter-site coordinations, while all thosewith the x coordinateof the outmost oxygen of the chain more rotated B-chains resembleroom-temperature and the x coordinateof the other basaloxygens. The orthopyroxene. In orthoferrosilite the coordination amounts of warping in enstatite (Morimoto and o switch occurswhen O3B-O3B-O3B = l5l (Suenoel Koto, 1969), anthophyllite (Finger, 1970), jim- al., 1976). There also seemsto be a tendency for thompsonite, and chesteriteare shown in Table 9. orthopyriboleswith wider chainsto have smaller ro- The chainsof enstatiteare not so warped as those of tation angles,and therefore the higher temperature the wider chain silicates,and the double B-chainsof outer M-site coordination. anthophyllite and chesteriteare more warped than any of the singleor triple B-chains. Polytype stabilities The distancesbetween basal oxygen layers in the N.Iuchhas been made of the importance of outer double- and triple-chain pyribolesare roughly equiv- M-site coordination to pyribole stability ranges.Pa- alent to the inter-layerbasal oxygen distancesof talc pike et al. (1973)suggest that pyroxeneM2 coordina- (Rayner and Brown, 1973),except for near the chain tion controls the ortho-clino transformation, and edges,where bonding within the adjacent M octa- Ihat P2'/c pyroxenesform metastablybecause their hedra pulls the chains together.If the warping were M2 sites have the low-temperatureortho coordina- the result of octahedral-tetrahedralmisfit, as in the tion. Smyth (1973) concurs with this notion, and antigorites, one would expect the triple-chain basal further suggeststhat the inability of the Pbca ortho- oxygen layersof jimthompsonite, clinojimthompson- pyroxenestructure to attain a higher symmetry con- ite, and chesterite to be uniformly bent. Instead, figuration reducesits stability at high temperatures most of the warping occursin the outermosttetrahe- relative to the monoclinic structure, which trans- dron, and the interior mica-like parts of the basal forms from P2r/c to the higher C2/c symmetry.sThe layer are quite flat. outer-site coordination must exercisesome control over the stabilitiesof the various polytypes,because Lamellae orientation pyriboles with large cations occupying these sites All specimensof clinojimthompsoniteand the un- never occur in the orthorhombic forms. However, it named mineral that have been observed from the seemsunlikely that the relativepositions of the outer Chesterwall zone occur as (100) lamellaein jim- M cations and their most distant coordinating oxy- thompsonite and chesterite,respectively. This is the gens can completelycontrol the thermodynamic be- same orientation that is commonly seen in inter- havior of the magnesianpyriboles. Other factorsmay growths of orthorhombic and monoclinic pyroxenes also affect the relative stabilities. and amphiboles.The (100)ortho-clino orientationis The data of Suenoet al. (1976) and Smyth (1973), probably common to all the known pyribole groups which are shown on Figure 12, suggestthat A-chain becausenearest-neighbor coordinations at the bound- misfit in orthopyroxeneincreases with rising temper- ary betweentwo polytypes sharing (100) are nearly ature. It is possiblethat the regular octahedraof the the same as in the two structures away from the structureare unable to deform enough to absorbthis boundary; only second-nearest-neighborrelation- misfit, and that this inability is in part responsiblefor ships are upset. reconstruction to the C2/c structure, which has no x- chainsand thereforeno chain misfit constraints. Cation ordering Refinement of M-site Mg-Fe occupancies for jimthompsonite, clinojimthompsonite,and chesterite 3 Monoclinic C2/c pyroxenes must be considered to possess revealsthe sameordering pattern observedin the low- higher symmetry than orthorhombic Pbca ones, becausethe den- calcium pyroxenesand sity of symmetry elements is greater, and hence there are fewer amphiboles:Fe2+ is concen- atoms per asymmetric unit. Also the M-site coordination poly- trated in the outer, distorted sites,while Mg prefers hedra possesshigher point symmetry. the inner, more regular sites. This is an expected VEBLEN AND BURNHAM: BIOPYRIBOLES l07l result, as the outer M-site polyhedralgeometries and Table 9 Degreeof chain warping (A) in low-Ca orthopyriboles environmentsare similar for all the ferromagnesian pyriboles,when only the five oxygensnearest this site A chaln B chain are considered;the differencesin the sixth- and sev- enth-nearestoxygens, as previouslydiscussed, appear Ensta ti te 0.14(2) 0.24(2) 0"19 to have little consequencefor Mg-Fe ordering ten- Anthophylli te 0.29 (2) 0.4L(2) 0.35 dencies.The M4 site of triple-chainI-beams, which is JimthonPsoni te 0.26(2) 0.28(2) O.21 analogousto the M2 site of amphiboles,appears to Chesterite: T-chains 0.29(3) 0.33(4) 0.31 D-chains 0.29 (4) 0.45(4) O.'37 be slightly enriched in Mg compared with the other regular octahedralsites. simple or whether it entails formation of still more ordered intermediatestructures. The amphibole-mica transformation Severalimplications of the above reaction scheme deservecomment. Becausethe transformation from Many pyribole single crystals from Chester pro- pure double chains to pure triple chains proceeds duce streaks parallel to b* on precessionfilms, in- through an ordered mixed-chain intermediate,it is dicating some degreeof structural disorder in the b possiblethat pyroxene-amphibolereactions produce direction. Preliminary results from high-resolution alternating single-double chain phases.In addition, electron microscopy have shown that the streaking thesereactions are likely to be marked by structurally resultsfrom errors in the widths of silicatechains in disorderedstates. the ordered minerals and from regions with appar- It has not, of course, been determined whether ently random sequencesof double, triple, and wider theseminerals have stability fields of their own. Be- chains(Veblen et al., 1977).Work is now in progress causethey occur in considerablequantities (the new to fully characterizethe microstructures found in orthorhombic phases can exceed lmm in a and c pyriboles.from Chester. dimensionsand reachseveral hundred microns in the The X-ray resultson the new mineralsshow us the b direction), they must be "nearly stable," even if general pattern of the amphibole-mica reaction. they formed metastably.The question of stability is Rather than being a simple one-step reaction, the moot in any case, becausethe biopyriboles of the anthophyllite-talc transformation is, at least in the Chester wall zone include at least sevenintimately Chester wall zone, a complex multistage reaction. coexisting phases which are apparently chemically The following sequencecan be inferred from the pet- co-linearand may includethree compositional coinci- rographic, chemical, and diffraction data on the dences:this is not an equilibrium stateof affairs.The Chesterbiopyriboles. new pyriboles will probably never attain valuable 1. Anthophyllite becomes structurally disordered, petrogeneticindicator status. acquiring triple-chain structure at random intervals; It is not clear whether the reversereaction, from Fe2+ and Mg'+ diffuse out of the structure as H+ talc to anthophyllite,takes place by the reversemech- diffusesin. anism, but there is some evidencethat the path may 2. When a region of the structurehas a high enough be similar. Daw et al. (1972) found in a TEM study concentrationof triple chains,it beginsto order into that enstatiteformed by heating talc produced elec- chesteriteor its monoclinic analog. Becausethese tron diffraction patternsthat were heavilystreaked in structurescontain rigorously alternatingdouble- and the b direction. They also observed (010) planar triple-chain slabs, the transformation from a dis- "faults" in the enstatite, reminiscent of the (010) ordered chain sequenceto theseminerals must entail lamellar featuresobserved in the Chester pyriboles. considerablereconstruction. Greenwood's(1963) rate study on growth of enstatite 3. More triple chains form, in excessof the 1:l from talc further indicates that anthophyllite is an double:tripleratio found in chesterite.More structur- intermediate product, and other phasescould have ally disorderedstates are the result. been lurking in the run products disguisedas an- 4. When all the double chainshave beenconsumed, thophyllite or enstatite.In any case,mica to pyroxene the transformation to the orderedtriple-chain miner- reactionsappear to take place by a complex mecha- alsjimthompsonite and clinojimthompsoniteis com- nism, rather than a single step. plete. The inferred anthophyllite-talc reaction sequence 5. Talc grows at the expenseof jimthompsonite. It is points out major problems in biopyribole reactions. not yet known whether this growth mechanism is The intergroup reactionsare controlled by multistep, to12 VEBLEN AND BURNHAM: BIOPYRIBOLES

highly reconstructivemechanisms, at leastat meta- bergite,jadeite, spodumene, and ureyite. Am. Mineral., 58, 594- morphictemperatures and time scales.Kinetic diffi- 6r8. cultieshave already been recognized in experimental Cromer, D. T. and D. Liberman (1970)Relativistic calculation of studies: (1963) anomalous scattering factors for X-rays. .I. Chem. Phys., 53, Greenwood was unableto nucleate l 89l-l 898. anthophyllitefrom talc in the anthophyllitestability - and J. B. Mann (1968) X-ray scatteringfactors computed field,but anthophyllitenucleated easily in partsof the from numerical Hartree-Fock wave functions. Acta Crystallogr., enstatitestability range. A24.32t-324. Daw,J. D., P. S. Nicholsonand J. D. Embury(1972) ln- Conclusion homogenous dehydroxylation of talc. J. Am. Ceram. Soc., 55, 140-t51. It is hopedthat this paperwill encouragepetrolo- Drits, V. A., Yu. I. Goncharov, V A. Aleksandrova, V. E. gistsand mineralogists who dealwith biopyribolesto Khadzhi and A. L. Dmitrik (1974) New type of strip silicate. be cautiousin theirobservations. Chain-width fault- Kristallografiya, 19, l186-1 193. (transl. Sou. Phys. Crystallogr., ing may proveto becommonplace in all the biopyri- t9,737-741). (1976) bole groups, altering thermochemical and I. P. Khadzhi Formation conditions as well as and physico-chemicalconstitution of triple chain silicatewith physicalproperties. Structurally-orderedminerals in- [SLOru] radical. (in Russian) Izuestiya Akad. Nauk., Ser. Geol., termediatebetween the pyroxenes,amphiboles, and No.7,32-41. micasmay proveto be commonminor phases,but Finger, L. W. (1969) Determination of cation distribution by least- theiridentification requires a combinationof careful squares refinement of single-crystal X-ray data. Carnegie Inst. petrographic,X-ray, and chemical lYash. Year Book, 67, 216-217 . observationson - (1970) Refinement of the of an anthophyl- fine-grainedmaterial. This would perhaps come as no lite. Carnegie Inst. Wash. Year Book,68,283-288. surpriseto Johannsen(l9t l), who, in givingthe su- Ghose, S. (1965) Mg'z+-Fez+ order in an orthopyroxene, pergroupthe name "biopyribole," recognizedthat MgoesFerorsi2Ou. Z Kristallogr., 122,8l-99. they all look very muchalike. Gibbs, G. V. (1969) Crystal structure of protoamphibole. Mineral. Soc Am. Spec. Pap.,2, l0l-109. Acknowledgments Greenwood, H. J. (1963)The synthesisand stabilityof anthophyl- lite. J. Petrol., 4, 317-351. This paper bearsthe clear mark of JamesB. Thompson's theo- Guggenheim,S. and S. W. Bailey(1975) Refinement of the marga- retical work on biopyriboles.We thank hirn warmly for our dis- rite structure in subgroup symmetry. Am. Mineral., 60, 1023- cussionsand for his shared insights. We also thank James F. t029. Hays, John B. Brady, Richard Sanford, and Kenneth Shay for Hanscom, R. H. (1973) The Crystal Chemistry and Polymorphism useful conversations.This work representspart ofa Harvard Uni- of Chloritoid. Ph.D. Thesis, Harvard University, Cambridge, versity Ph.D. thesis(Veblen, 1976).The final versionof the paper Massachusetts. was improved by thorough reviews from James J. Papike and Johannsen,A. (l9l I ) Petrographic terms for field use.J. Geol., l,9, Cornelis Klein. Financial support for this researchwas provided 3t7-322. by NSF grant GA-41415 to Charles W. Burnham. Katscher,H. and F. Liebau (1965) Uber die Kristallstruktur von BarSiaO", ein Silikat mit DreifachkelLen. Naturv)issenschaften, References r8.5t2-5t3. Kosoi, A. L., L. A. Malkova and V. A. Frank-Kamenetskii(1974) Bailey, S. W (1966) The status of clay mineral st.ructures.Proc Crystal-chemical characteristicsof rhombic pyroxenes. Kristal- Nat. Conf. Clays and Clay Minerals, 14, l-23. lografya, 19,282-288. (transl. Sou. Phys. Crystallogr., 19, 17l- - (1975) Cation ordering and pseudosymmetryin layer sili- t74). ca|es.Am. Mineral.,60, 175-187. I-iebau, F. (1972) crystal chemistry. In K. H. Wedepohl, -, G. W. Brindley, W. D. Keller, D. R. Wones and J. V. Ed., Handbook of Geochemistry,Vol. lI-2. Springer-Verlag, Ber- Smith (1967) AGI Short Course Lecture Notes-Layer Silicates. lin. American Geological Institute, Washington. McCauley,J. W. and R. E. Newnham (1971)Origin and prediction Brown, G. E. and G. V Gibbs (1970)Stereochemistry and order- of ditrigonal distortions in micas.Am. Mineral.,56,1626-1638. ing in the tetrahedral portion of silicates Am. Mineral., 55, Morimoto, N. and K. Koto (1969)The crystalstructure of ortho- 1587-t607. enstatite. Z Krisrallogr, 129, 65-83. -, C. T. Prewitt, J J. Papike and S. Sueno (1972) A com- -, Y. Nakajima, Y. Syono, S. Akimoto and Y. Marsui (1975) parison of the structures of low and high pigeonite. J Geophys. Crystal structures of pyroxene-type ZnSiO, and ZnMgSirOu. Res.. 77. 5778-5789. Acta Crystallogr., 83l, 104l-1049 Burnham, C. W. (1967) Ferrosilite. Camegie Inst. Wash. Year -,C. T. Prewitt,S. Suenoand M. Cameron(1973) Pyroxenes: Book. 65. 285-290. comparisons of real and ideal structural topologies. Z. Kristal- -, Y. Ohashi, S. S. Hafner and D. Virgo (1971) Cation logr., 138,254-273. distributionand atomic thermal vibrationsin an iron-rich ortho- - 2nd M. Ross (1970) Gedrites: crystal structures and intra- pyroxene. Am. Mineral., J6, 850-876. crystaffine cation distributions. Am. Mineral., 55, 1945-1972. Cameron, M., S. Sueno, C. T. Prewitt and J. J. Papike (1973) Pauling, L. (1948) The Nature of the Chemical Bond. CornellUni- High-temperature crystal chemistry of acmite, diopside, heden- versity Press,lthaca, New York. VEBLEN AND BURNHAM: BIOPYRIBOLES lo73

Rayner,J. H. andG. Brown(1973) The crystalstructure of talc. - (1978)Biopyriboles and polysomaticseries. Am' Mineral', Clays and Clay Minerals, 21, 103-114. 63,239-249. Smith,J. V. (1969)Crystal structure and stabilityof the MgSiO' Veblen, D. R. (1976) Triple- and Mixed-Chain Biopyribolesfrom polymorphs:physical properties and phaserelations of Mg,Fe Chester.Vermont. Ph.D. Thesis, Harvard University, Cam- pyroxenes.Mineral. Soc. Am. Spec.Pap., 2,3-29. bridge,Massachusetts' Smyth,J. R. (1971)Protoenstatite: a crystal-structure refinement - 2nj C. W. Burnham(1975) Triple-chain biopyriboles: at I100'C. Z. Kristallogr.,134,262-274. newly discoveredintermediate products of the retrogradean- - (1973)An orthopyroxenestructure up to 850oC.Am. Min- thophyllite-talctransformation, Chester, Vt' (abstr')' franr' eral.. 58.636-648. Am. GeophYs.Union, 56, l0'16. - (1974)Low orthopyroxenefrom a lunardeep crustal rock: a - sni - (19'16)Biopyriboles from Chester,Vermont: ' new pyroxenepolymorph of spacegroup P2rca. Geophys. Res. the first mixed-chainsilicates (abstr.). Geol. Soc' Am. Abslracts Leu.,1,27-29. with Programs,8, I 153. - 3nd C. W. Burnham(1972) The crystalstructures of high - 2ni - (1978)New biopyribolesfrom Chester'Ver- andfow clinohypersthene. Earth Planet. Sci. Len., 14,183-189. mont: L Descriptivemineralogy. Am' Mineral',6-t, 1000-1009' - and J. lto (1977)The synthesisand crystalstructure of a -, P. R. Buseckand C. W. Burnham(1977) Asbestiform magnesium-lithium-scandiumprotopyroxene. Am. Mineral., chainsilicates: new minerals and structuralgroups' Science, l98' 62. t252-1257. 359-365. Sueno,S., M. Cameron,J. J. Papikeand C. T. Prewitt(1973) The Warren,B. E. (1929)The structureof tremoliteHrCarMgu(SiOs)g' high temperaturecrystal chemistry of tremolite.Am. Mineral., Z. Kristallogr., 72, 42-57. s8.649-664. - and D. I. Modell (1930)The structureof anthophyllite and C. T. Prewitt (1976)Orthoferrosilite: high- H,Mg"(SiOrL.Z. Kristallogr.,75' 16l-178' temperaturecrystal chemistry. Am. Mineral.,6l;38-53. Takeda,H. (1972)Crystallographic studies of coexistingaluminan orthopyroxeneand augiteof high-pressureorigin. J. Geophys. Res.,77,5798-581l. Thompson,J. 8., Jr. (1970)Geometrical possibilities for amphi- Manuscriptreceiued, October 13,1977; accepted bofe structures:m odel biopyrib oles. A m. M ineral., 5 5, 292-293. for publication,MaY i,6, 1978'