Ameiran Mineralogist, Volume 60, pages 749-757, 1975

Microstructure and Crystallinity of Gem Opals

I. V. SlNnrns

CSIRO Diuision of Tribophysics, Uniuersity of Melbourne,Parkuille, Victoria, 3052. Australia

Abstract

Gem opals, from volcanic host rocks from a variety of sources, have been examined byelectron microscopy and diffraction. They are generally a mixture of amorphous and crystal-line silica, the extent of crystallinity varying between samples from different localities. Thecrystalline phase in some samples has been identified as tridymite. Their microstructures arecompared with those of gem opals from deposits in sedimentary rocks, and with specimensheated in the laboratory. Changes in morphology produced by sintering occur at about400'C, and crystall ization at about 1100'C. Both tridymite and cristobalite were identif ied inmaterial recrystallized by heating.

Introduction

The major sources of the world's gem opals are theAustralian fields of Andamooka, Coober Pedy, andLightning Ridge. Here, the opal occurs in sedimen-tary host rocks, where it was deposited in cavities byconcentration of an aqueous solution of sil ica, form-ing first a gel and finally particles of silica generally0.1-0.5 pm in diameter (Darragh et al, 1966).In thistype of gem-quality opal the particles are uniform insize and form a three-dimensional optical diffractiongrating which produces the play of color (Sanders,1968). The much more common potch opal accom-panying the gem quality material is similar in struc-ture, but the particles are not uniform in size or shape(Sanders, 1964). Because the particles consist ofseveral (generally ( 5) concentric shells, Darragh et al(1966) suggested that they were formed by aggrega-tion in a gel of smaller, primary particles severalhundred Angstroms in diameter.

Precious opal also occurs in smaller quantit ies inmany other places in the world, but mostly in con-junction with rocks of volcanic origin rather than insedimentary host rocks. The most extensive and bestknown deposits occur in Mexico, where opal is minedcommercially. Typically it is found in vesicles insolidif ied lavas, and because it exhibits a weaker playof color than that of opal from the sedimentary fieldsin Australia, it can generally be distinguished by eye.Opal of l ike origin and similar in appearance to Mex-ican opal is also found in Australia near the eastcoast, but in smaller quantities than on the sedimen-tary fields. Electron microscopy showed that the

diffracting structure of some of these opals is less dis-tinct than that of opals from sedimentary host rocks,but the diffracting arrangement of particles can oftenbe revealed only by etching the fracture surfaces andexamining replicas (Darragh and Sanders, 1969;Sanders and Darragh, l97l).

This paper compares the structure and crystall inityof precious opal from these less common volcanicdeposits with that of gem opal from the Australiansedimentary fields. The two types will be dis-tinguished as "volcanic" or "sedimentary" opal. Forcompleteness comparison is also made with somesynthetic samples.

An extensive examination by X-ray diffraction ofopaline silicas has established that gem opals are inthe least crystalline group and give a diffuse diffrac-tion pattern of broad rings corresponding to latticespacings of 4.1,2.0, 1.5, and 1.2 A (Jones and Segni t ,197 l). Electron microscopy has many advantagesover X rays for examining materials of poor crystal-l inity, such as opal. First, images can be obtainedfrom thin fragments, showing the shape and distribu-tion of any crystalline components. Second, diffrac-tion patterns can be obtained and used to identifyphases, and because of the shorter wavelength of theelectrons, l ine broadening from small crystals is muchless than for X rays. Third, single crystal patterns canbe obtained from individual crystals less than I pm inSIZE.

It seems possible that when crystall inity occurs ingem opal, it may have been caused by a temperaturerise produced by a subsequent flow of lava over theopal deposit. Samples of specimens of each type of

749

'ts0J, V, SANDERS

opal were therefore examined by transmissionelectron microscopy after they had been heated in acontrolled manner to temperatures up to l200oC.

Experimental

As far as possible, specimens were from identifiedlocalities, and frequently with the host rock attached.The kindness of many people in providing overseasspecimens is gratefully acknowledged.

Three different synthetic samples were examined.Two were made by the C.S.I .R.O. Divis ion of Ap-pl ied Mineralogy (Ll , L2 in Table 3); Ll was madeby the concentration of pure silicic acid in aqueoussolution at about 80oC, and the subsequent separa-tion to produce mono-dispersed spheres of silica; forL2, spheres of silica were made from tetraethylortho-silicate dissolved in alcohol and reacted with water(Stdber, Fink, and Bohn, 1968). In both cases thespheres were allowed to settle, dehydrate, and consol-idate. The third sample, L3, was a commercial sam-ple, kindly provided by P. Gilsen.

A small (- 10 mg) typical piece was broken offeachspecimen and ground to a powder in a mortar, firstdry and then in acetone. Fragments were collected oncarbon-coated grids. In annealing experiments aspecimen was broken into pieces (- I gm) which were

heated in an open ceramic boat in a tube furnace at aset of controlled temperatures. Small pieces were sub-sequently broken off and ground to fragments whichwere mounted for examination in the electronmlcroscope.

Specimens were examined in a Philips EM 200electron microscope at 100 kV. Diffraction patternswere obtained from individual fragments about I irmin size, isolated by a selector aperture.

Results

Sedimentary Fields

Microstructure. When opals fracture, the breakgenerally occurs through rather than between thesil ica particles, so that thin fragments formed bygrinding consist of sections of the particles and thenaturally formed cavities between them. Figure lashows how the appearance of fragments in anef ectron microscope at lower magnifications isdominated by a regular array of cusp-shaped cavities;sections of the spherical particles can also be dis-tinguished. No internal structure is visible in thinedges of these fragments at higher magnifications(Fig. lb); the small variation in contrast or graininessis about the same as that from the supporting carbon

FIc. l. Transmission electron micrographs of fragments of gem opals from sedimentary hosts: (a) An-damooka, 35,000 X: insert , d i f f ract ion pat tern consist ing of a strong halo at about a A; (b) detai l of the

thin edge of a fragment from Lightning Ridge, 400,000 X

MICROSTRUCTURE AND CRYSTALLINITY OF GEM OPALS

fi lm which always shows some random variation ofelectron density in slightly underfocussed images.Crystallinity. Although it had been establishedpreviously that opal from sedimentary fields isamorphous to electrons (Jones, Sanders and Segnit,1964), samples from a number of these fields were re-examined, and samples from a few additional fieldsincluded. The diffraction patterns from all samples(Table l) consisted mainly of a diffuse halo centeredon a spacing of about 4.1 A (Fig. I a insert), any otherhaloes being concealed by the diffuse rings of thecarbon supporting film (Jones, Milne, and Sanders,1966); these opals are therefore confirmed to beamorphous.

Deposits in Volcanic Rocks

Microstructure. The appearance in the electronmicroscope of opal from volcanic hosts (Table 2)depended upon i ts or igin. Two samples fromAustralian deposits (VI, Rocky Bridge Creek, andV2, Tooraweenah) and one sample from Czerwer-nitza (V3) appeared similar to those from sedimen-tary deposits. In all other samples no cusp-shapedholes were detected, and there was no evidence atcomparable magnifications (Fig. 2a) of the diffractingarray of particles seen in sedimentary opals (Fig. la).At higher magnifications a microstructure of smallparticles could be resolved to varying extents. Theseparticles, about 100 A in size, linked together, areshown in Figure 2b. In other samples, although theindividual particles could not be resolved, the frag-ments nevertheless appeared porous (Fig. 2c) and areso described in Table 2.Crystallinity. The diffraction patterns (Fig. 3)varied from a single halo (d: 4.1A) as for "sedimen-tary" opal, through sharper continuous rings, tospotty rings in which single crystal nets of spots couldbe distinguished. Diffraction patterns like these,taken from single fragments of about the same size,were used to assess the extent of crystallinity in sam-ples from various sources (Table 2).

The most significant feature of these patterns is theinnermost ring, corresponding to the halo in Figure3a. It consisted sometimes of a single ring whosesharpness depended upon the sample (Fig. 3b), or inthe most crystalline samples it consisted of at leasttwo resolved rings, whose positions were usuallydefined by spots (Fig. 3c). In the latter case thepresence of tridymite is immediately suggested, andmeasurement confirmed the identification.

Figure 4 shows the positions of the most intensereflections expected from cristobalite and tridymite

TABLE l. Opal Associated with Sedimentary Host Rock

Local i ty CountryDegree of

Crystal l izat ion

S 1

> z

S 3

S4

S5

S6

J /

S E

Andamooka

Coober Pedy

Wh i t e C l i f f s

Kal goor l ie

Lightning Ridge

Grawin

Qui lp ie

2

SouthAustra l ia

SouthAustra l ia

N . S . W . ,Austra l ia

W . A . ,Aust ra l ia

N . S . W . ,Austra l ia

N . S . W . ,Austra l ia

Queens I and ,Austra l ia

Braz i I

None

None

None

None

None

None

None

None

(Dollase, 1965; Frondel,1962), and indicates the diffi-culty of distinguishing a mixture of cristobalite andtridymite from pure tridymite in the patterns ofdiffuse rings. Careful measurement of patterns likethat in Figure 4, in which the innermost ring was notresolved into components, gave D : 4'12 t 0'03 andd = 2.50 + 0.01 A for the two clearest rings. Thisdoes not distinguish the material as cristobalite ratherthan tridymite, but the width of the innermost ringsuggests that the crystalline phase is tridymite. An-nealing experiments, to be described later, supportthis interpretation.

The diffraction patterns from the most highly

T/'srE 2. Opal Associated with Volcanic Host Rocks

L o c a l i t y CountryInternal Degree of

S t r u c t u r e C r y s t a l l i n i t Y

VI Rocky BridgeCreek

v2 Tooraweenal

V3 Czeruen i tza IV4 SpencerVS l\'le I an i e

V6 Mt, Bougrom

v 7 ?

V8 V i rg in Va l ley

V9 near Redlands

V10 Teven

v l 1 ?v72 ?

V l3 Lead P ipeSprings

V14 Czeryen i tza I I

N . S . W . ,Aust ral i a

N . S . W . ,

A u s t r a l i a

I d a h o , U . S . A .

Q u e e n s l a n d ,Aus t raI ia

Queens land ,Aus t ral ia

Indonesi a

Nevada,

u . s . A .

u . s . A .

N . S . W . ,Aus t ra l ia

MexicoHonduras

u . s . A .Hungary

None None

None None

None NoneP o r o u s S l i g h t

t oo - r so R s l i gh tprlnarles

r00 A S l ish tpranaTles

Porous S l igh t

so- roo R s l i sh tprrnarles

roo .R ModerateprlnaTaes

Porous Moderateroo E Ex tens iveprlnarles

70-150 X Ex t .ens ivepr lna t les

5o-1oo X. Ex tens ive

Prrnatles

J. V. SANDERS

FIc. 2. Microstructure of volcanic gem opals composed of l0 nm primary particles: (a) Melanie,Queensland (v4),35,000 x: (b) Melanie, Queensland (v4),450,000 x; (c) Mexico (v l0) ,300,000 x.

crystall ine samples (e.g., Y12, Lead Pipe Springs,Cal i forn ia, U.S.A.) conta ined st reaks passingthrough the first ring of the group of three innermosttridymite rings (Fig. 3d). Some suitable fragments

were selected and tilted in the electron mciroscope togive cssentially single crystal patterns which weresimple reciprocal lattice nets. From these it was es-tablished that the six patterns analyzed were consis-

FIc 3 Examples of the types of diffraction patterns given by gem opal from volcanic host rocks.Specimen localities are: (a) Rocky Bridge creek, N.S.w. (vI); (b) Mt Bougrom, eueensland (v5); (c)Tevern, Queensland (V9); (d) California (Vl2); the diffraction pattern in (d) is reproduced ar a lower

magnification to show the diffraction streaks.

MICROSTRUCTURE AND CRYSTALLINITY OF GEM OPALS 753

tent with tridymite and not with cristobalite. Thestreaks passed through hll (h : 2n) tridymite reflec-tions in the c* direction.Morphology of the Crystalline Component. Bymeans of the dark field technique in which an imageis formed by transmitting diffracted electronsthrough the objective aperture, it is possible to dis-tinguish those parts of a fragment contributing to thecrystall ine component of the diffraction patterns. Inthis way it was established that in general only smallparts of fragments were contributing to the sharprings. Having identified such areas at first by thedark-field technique, one can subsequently recog-nize them in normal bright-field images where theyappear as dark patches in a more homogeneous back-ground. They are marked by arrows in Figure 5.

ln samples j udged to be "slightly" to "moderately"crystall ine, and giving diffraction patterns as inFigure 3b, most fragments contained irregularly-shaped crystals with maximum dimensions about 50-200 A. Figures 5a and b show examples of suchcrystall ine patches in a matrix of the primary parti-cles. Generally only a small proportion of each frag-ment was crystall ine. Where crystall inity was moreextensive, some crystals took the form of needlesabout 100 A wide and about 1000 A long, sometimesin bundles (Fig. 5c). The most extensive crystall iza-tion yet observed was found in a sample of gem opal

Frc 4. Diffraction pattern of a sample from Mt. Bougrom,

Queensland (V5) together with 111 and200 rings from aluminium.

The most intense lines from cristobalite (C0) and tridymite (Za)

are marked

known to be from Mexico, but otherwise uniden-

tif ied. It appeared to have been almost completely

converted to crystall ine needles, of more or less

uniform size in random directions and so forming a

cr iss-cross network throughout the sample.Micrographs (Fig. 6) of a surface extraction replicashow its structure, at low magnification (Fig. 6a), at

intermediate magnification (Fig. 6b) showing the

network of needle-shaped crystals, and at high

magniflcation (Fig. 6c) showing the shape and inter-

nal structure of individual needles. These crystals aregood examples of the structure and shape of well-

developed crystall inity in opal generally.

Intense irradiation with electrons in the electron

microscope destroyed the crystall inity, although the

shape of the crystall ine area was retained. For this

reason, care is necessary in studying these materials

Frc. 5. Examples of crystals within a matrix of otherwise amorphous silica in fragments of volcanic gem

opal. (a) "slightly" crystalline, Mt. Bougrom, 350,000 X; (b) "moderately" crystalline, Honduras,600,000

X; (c) "extensively" crystal l ine, Honduras, 250,000 X.

754 J. V. SANDERS

Ftc 6 Anextensivelycrystal l inegemopal f romMexico:(a)surfaceextract ionrepl ica; thedarkerpatches,about I mm in size, are the regularly arranged particles which make up the diffracting array; 30,000 X. (b)part of area shown in (a); 50,000 X. (c) Fragment showing almost complete conversion to needles, and

faults within the crystals, 140,000 X.

at high magnification to ensure that no changes areproduced by the electron beam.

Synthetic Material

In three different synthetic samples no crystallinitywas detected, and no microstructure could be seen bytransmission electron microscopy of thin edges offragments (Table 3). In these respects the microstruc-ture of the synthetic material is similar to that of opalfrom sedimentary deposits.

Annealing

Heat ing caused crysta l l izat ion and internalmorphological changes due to sintering in bothvolcanic and sedimentary opals. In all cases sinteringpreceded the onset of crystall ization by about 700'C.Sintering. Opal from both sedimentary andvolcanic deposits contains natural voids. In theformer they are the regularly arranged cavities left

Trsle 3. Synthetic Opal

L o c a l i t y CountryDegree of

Crystal 1 i zat ion

between the spherical particles and are thereforecusp-shaped holes with a maximum dimension ofabout 0.1 pm (Fig: la). In the "volcanic" opals theyare the much smaller, irregular spaces between theprimary particles (Fig. 2). Sintering by heat changesthe shapes of the cavities into spherical voids, so thatone then sees circular pores in an otherwisefeatureless matrix. In both types of opal, sintering issimply detected by a close examination of the shapesof the cavities, which change from cusps to roundedholes. A set of specimens from each of the threedifferent Australian sedimentary fields of LightningRidge, Andamooka, and White Cliffs, was heated forl0 minutes at a variety of temperatures up to I 100"C.For each set sintering became appreciable in thetemperature range 350oC < T < 480'C. Although themajority of fragments which were viewed containedround holes in sintered samples, some fragmentsshowed no signs of sintering; this indicates either aninhomogeneity in the sample, or that the time was in-sufficient for temperature equilibrium to have beenattained. In gem quality sedimentary specimens thatexhibit diffraction colors, heating to 600'C or 700oCchanged the shapes ofthe holes but did not alter theirspatial arrangement, so that they still formed aregular array (Fig. 7c; cf Fig.la). This is in keepingwith the observation that heating up to about 1000"C

L 1L 2L J

CSIRO ICS IRO I IGi I sen

Austra l iaAustra l iaSwit zer land

NoneNoneNone

generally does not destroy the color in opal, showingthat the diffracting array is retained.

The onset of sintering is less easily detected in sam-ples from volcanic sources, but it was found to be inthe range 400'C < T < 600"C. The change to roundcavities of irregular size is shown in Figures 7a, b.Crystall ization. Crystall ization is detected bydiffraction, just as for natural material. In opal fromsedimentary fields, the init ially amorphous materialwas completely crystallized after 3 hours at l200oC togive irregularly shaped grains, about I pm diam., ofeither cristobalite or tridymite. Partial crystallization(e.5., V, hour, l200oC) produced similar grains quiteunlike the needle-shaped crystals in naturally occur-ring opal, in an otherwise amorphous matrix. Diffrac-tion patterns from a completely recrystallized sample(Fig. 8), with many fragments in the diffraction field,look very similar to the patterns from well crystal-l ized natural specimens (Fig. 3c). In particular, thering from spacings near 4 A is well resolved into com-ponents. With this sample it was possible to selectfragments which were single crystals and by orientingthem in the electron microscope to obtain simplereciprocal lattice sections. Positive identification ofthe phase is then possible; it was found that the lowtemperature form of cristobalite (Dollase, 1965)

755

predominated but some patterns from high tridymitewere also identified, (of fourteen patterns analyzed,twelve were from cristobalite and two fromtridymite.) With this knowledge it is possible to seethat the ring patterns of Figures 8 and 3c representpatterns from these two phases, super-imposed; con-sequently, the group of rings near 4 A spacings con-sists of a central component containing many spotscorresponding to a high concentration ofcristobalite,and less densely populated outer components fromsmaller concentrations of tridymite.

The crystals of cristobalite were twinned and con-tained faults on planes parallel to (l0l). However,there was no evidence of an intergrowth of

Frc. 8 Diffraction pattern from initially amorphous opal fromAndamooka (Sl) after being heated to 1200'C for 3 hours. Thisdiffraction pattern comes from an area containing many frag-ments.

MICROSTRUCTURE AND CRYSTALLINITY OF GEM OPALS

Ftc 7. Morphological changes produced by heating amorphous gem opals: (a) before heating samplel iom Mt. Bougrom, Queensland (V5),300,000 X; (b) same sample af ter being heated for 4 hours at800'C,300,000 X; (c) opal f rom White Cl i f fs , N.S.W (S3) af ter being heated for l0 min at 700'C,40,000 x.

756 J. V. SANDERS

cristobalite and tridymite. The patterns from singlefragments of tridymite contained no reflections con-sistent with intergrown cristobalite, and vice-versa.

Discussion

Within the limited number of localities examinedhere, "sedimentary" gem opals consistently exhibitno microstructure in transmission electron micro-scope (Teu) images at high magnification, to thelimit of the present resolution (-4 A). Their electrondiffraction patterns were all diffuse, suggesting thatsuch opals are amorphous. The synthetic sampleswere also structureless at this scale, and amorphous.

Some opal from volcanic rocks has a similar struc-ture, but more often it is partly crystalline and the ex-tent of crystallinity varies with its origin. Further-more, a microstructure of primary non-crystallineparticles, about 100 A in diameter, can frequently bedetected in these specimens. This structure is consis-tent with the appearance of replicas of fracture sur-faces of these types of opal (Darragh and Sanders,1969) in which a rough texture on a 100 A scale isvisible. Similar structures have been recognized inphytoliths (silica particles within plants), tabashir(silica deposited inside bamboo), and in silica gel(Jones et al, 1966). In these cases the silica is certainlydeposited at ambient temperatures from an aqueoussolution. It therefore seems likely that the primaryparticles in opal in volcanic rocks were formed undersimilar ambient conditions.

It is surprising that the structure of concentricshells in spheres in opal from sedimentary deposits(Daragh et al, 1966) is not apparent in thin fragmentsexamined by direct Tru. This implies that betweenthe primary particles there is infilling by silica whichis indistinguishable by Tru but which has sufficientlydifferent solubility in HF for the primary particles tobe made visible by etching. If the shell width repre-sents the diameter of the primary particles, this isabout 500 A, and is therefore much larger than thesize of primary particles in volcanic opals.

Many of the samples from volcanic deposits con-tained crystals imbedded in the otherwise unchangedamorphous matr ix. The appearance of thesecrystals-and the way they have grown within thematrix of primary particles to sizes greater than thatof the particles-is similar to that of crystals formedby heat ing compacts of amorphous mater ialsproduced by precipitation from aqueous solutions,such as AIOOH, ZrOz, MoOg. These materials arealso composed of primary particles a few hundredAngstroms in diameter, and when heated, they

recrystall ize before they sinter; crystall izationproceeds from particle to particle to produce crystalslarger than the primary particles (Garrett andSanders, unpublished).

The relative stabilities of the phases of silica havebeen considered recently by Jones and Segnit (1972).

Quartz is the phase of silica thermodynamicallystable at ambient temperatures; the transition totridymite occurs at 876oC and the tridymite tocristobalite at 1470oC. The absence of reflectionswith a lattice spacing of 3.4 A, a diagnostic forquartz, shows that quattz is not present in thespecimens examined. The next most l ikely phase istridymite, and its presence in volcanic opals has beenestablished here. However, Wahl, Grim, and Graf(1961) using X rays, found that amorphous opalcrystall ized to cristobalite above 1000'C, and theydetected no tridymite in their experiments on heatedquartz and opal. Also Mizutani (1966) detected notridymite when amorphous silica was heated at 300'Cin aqueous KOH under pressure, f inding cristobalitefirst and then quartz. It seems likely that the tridy-mite crystals are too small to be distinguished byX rays, particularly in the presence of small cristo-balite crystals, but that they can be recognized inelectron diffraction patterns.

The essential difference between cristobalite andtridymite is the stacking sequence of hexagonallylinked tetrahedra of SiOo, the sequence being cubic incristobalite and hexagonal in tridymite (Frondel,1962). There was no evidence for an intergrowth ofcristobalite and tridymite as might be expected if thestacking sequence changed from cubic to hexagonalwithin the one grain.

The co-existence of separate crystals of cristobaliteand tridymite in samples heated in the laboratorysuggests that crystallization was nucleated in the ran-dom network of l inked SiO, tetrahedra at placeswhere l inkages naturally occurred in the planar ar-rangement which would be appropriate for the for-mation of cristobalite or tridymite. Along withfaulting and microtwinning, once a stacking sequencehad been established, it was retained in furthergrowth and determined whether the crystal becametridymite or cristobalite.

Bossi er al (1973) found that pure amorphous sil icadid not sinter below I l00oC, and that the onset ofsintering was lowered to about 600'C by the additionof 1-3 percent NaOH. Australian opals have beenfound to contain generally up to 0.5 percent Na(Bayliss and Males, 1965). This suggests that im-purities control the sintering kinetics in natural opals.

MICROSTRUCTURE AND CRYSTALLINITY OF GEM OPALS I ) I

The lower temperatures observed here for the onsetof sintering may be due in part to the nature of theimpurit ies, and in part to the high sensitivity of Tnufor detecting the morphological changes produced bysintering.

Several conclusions can be drawn from the obser-vation that opals heated in the laboratory sinter at amuch lower temperature than that required forcrystallization, whereas there is no sign of sintering inopals containing naturally-formed crystals. First, thisseems to preclude the possibility that crystallizationin natural volcanic opals was produced thermally bya flow of lava on top of the opal deposit after its for-mation. Second, it suggests that the mechanism ofnatural crystallization was quite different from thatof laboratory heated samples.

The process of sintering is one of molecular re-arrangement. By contrast, to form crystals oftridymite or cristobalite, which contain planar hex-agonal Si-O networks, bonds must be broken in therandom chains in the amorphous network. This iswhy in anneal ing exper iments much h ighertemperatures are required for crystallization than forsintering. But when crystals are observed in naturalvolcanic opals, the matrix is not sintered, showingthat the crystallization has taken place by a lowtemperature process. In natural crystall ization, im-purities may aid the rupturing of bonds, in a catalyticmanner. The temperature being low, the impuritycannot diffuse away from the site of nucleation andso crystall ization can continue there. On this basis theconcentration of crystals should depend on the con-centration of the impurity and, because the size of thecrystals is perhaps indicative of the time the processhas been operating, size should be a measure of theage of the material. If this is so, then the opal fromthe sedimentary fields contains no crystals eitherbecause it is younger than that from volcanic rocks,or because it does not contain the type of impuritynecessary to catalyze crystallization.

Acknowledgments

It is a pleasure to acknowledge the continued co-operation ofP. J. Darragh and J. Perdrix, and their assistance in the annealingexperiments. I am grateful to many people for opal samples, inparticular to T. R. Barbour for a fine set of specimens fromHonduras.

ReferencesBnylrss, P., nNo P. A. Melrs (1965) The mineralogical similarity

of precious and common opal from Australia. Mineral. Mag.35,429-431.

Bosst , A. , G. Lrorarrr , E Monnrrr , eNo N. GronolNo (1973)

Morphological and structural effects of NaOH added to silica. "/.Ma t . Sc i . 8 , l l 0 l - 1109 .

Dennncu, P. J. , A. J. GesrrN, B. C. TERRELL, eNo J. V. SeNorr .s(1966) Origin of precious opal. Nature,209, 13-16.

eNo J. V. SlNoens (1969) Volcanic gem opal examinedby electron microscopy. Awt. Gemmolog,J, 10, 5-8.

Dot-t-lse, W. A. (1965) Reinvestigation of the structure of lowcristobalite Z. Kristallogr l2l, 369-377.

FnoNosr-, C. (1962) Dana's System of Mineralogy, Vol. III. JohnWi l ey . New Yo rk , p . 261 .

JoNrs, J. B. , J . V SnNosRs, AND E. R. SEcNrr (1964) Structure ofopal. Nature, 204, 990-991.

- , app f f t SrcNrr (1971) The Nature of opal . LNomenclature and constitutent phases. J. Geol. Soc. Aust.18,57-68

-, AND - (1972) Genesis of cristobalite and tridymite atlow temperatures. J Geol. Soc. Aust. 18, 419422.

JoNss, L. H P. , A. A. Mrr-Nr eNo J, V. SrNnrns (1966) Tabashir :An opal of plant origin. Science,151,464-466.

MtzurnNI, S. (1966) Transformation of silica under hydrothermalconditions. J. Earth Sci. 14, 56-88.

SnNorr.s, J. V. (1964) The colour of precious opal. Nature,204,I l 5 l - l 1 5 3 .

- (1968) Diffraction of light by opals. Acta Crystallogr A24,427-434.

--, AND P J. Dlnucu (1971) The Microstructure of preciousopal. Mineral. Rec. 2, 261-268.

Srdsrn, U. , A. FrNK, rNo E. Bour.r (1968) Control led growth ofmonodispersed silica spheres in the micron size range. J. ColloidInterface Sci 26, 62-69.

Wrul , F. M., R. E. Gnru e.No R. B. Gur (1961) Phase t ransfor-mations in silica as examined by continuous X-ray diffraction.Am. Mineral. 46, 196-208.

Manuscript receiued, September 15, 1974; accepted

for publication, April 30, 1975