The superstructure of plagioclase feldspars

A modulation function representing the position and density of (Na, Ca) atoms in the superstructure of the e-plagioclase has been derived from the average structures of different plagioclase and a general modulation theory. Based on this function the superstructure of bytownite (An₇₃) has been studied with the single crystal X-ray method. The cell dimensions by Megaw's axes are a=7.946(3)A, b=67.09(2)A, c=12.236(4)A, α=39.03(1)°, β=45.63(1)° and γ=59.63(1)°. Z=18(Na, Ca) Al(Al, Si)Si₂O₈. The initial phase factor of the modulation function for bytownite has been obtained from the intensity data of the satellite reflections. This modulation function indicates a coherent small-scale alternation of the Na-rich and Ca-rich bands in the superstructure. This superstructure has been refined by applying the albite and anorthite structures to the Na-rich and Ca-rich bands, respectively. The change of the superstructure of the e-plagioclase due to the compositional change has been described based on the movements of the satellites in reciprocal space. The direction of the coherent small-scale intergrowth of the anorthite-like and albite-like bands is perpendicular to the t vector. The thickness of the intergrowth is 1/|t|. Both direction and thickness change regularly from An₇₅ to An₂₅.     

Thermal expansion of plagioclase feldspars

The volume thermal expansion coefficient and the anisotropy of thermal expansion were determined for nine natural feldspars with compositions, in terms of albite (NaAlSi₃O₈, Ab) and anorthite (CaAl₂Si₂O₈, An), of Ab₁₀₀, An₂₇Ab₇₃, An₃₅Ab₆₅, An₄₆Ab₅₄, An₆₀Ab₄₀, An₇₈Ab₂₂, An₈₉Ab₁₁, An₉₆Ab₄ and An₁₀₀ by high resolution powder diffraction with a synchrotron radiation source. Unit-cell parameters were determined from 124 powder patterns of each sample, collected over the temperature range 298–935 K. The volume thermal expansion coefficient of the samples determined by a linear fit of V/V ₀ = α(T − T ₀) varies with composition (X _An in mol %) as:

Constraints on the thermodynamic mixing properties of plagioclase feldspars

Taking account of the Cˉ1/Iˉ1 (Al/Si order/disorder) transformation at high temperatures in the albite-anorthite solid solution leads to a simple model for the mixing properties of the high structural state plagioclase feldspars. The disordered (Cˉ1) solid solution can be treated as ideal (constant activity coefficient) and, for anorthite-rich compositions, deviations from ideality can be ascribed to cation ordering. Values of the activity coefficient for anorthite in the Cˉ1 solid solution (γ _An ^Cˉ1 ) are then controlled by the free energy difference between Cˉ1 and Iˉ1 anorthite at the temperature (T) of interest according to the relation: ΔˉG _ord ^Iˉ1 ⇌Cˉ1 =RT ln γ _An ^Cˉ1 . If the Iˉ1⇌Cˉ1 transformation in pure anorthite is treated, to a first approximation, as first order and the enthalpy and entropy of ordering are taken as 3.7±0.6 kcal/mole (extrapolated from calorimetric data) and 1.4–2.2 cal/mole (using an equilibrium order/disorder temperature for An₁₀₀ of 2,000–2,250 K), a crude estimate of γ _An ^Cˉ1 for all temperatures can be made. The activity coefficient of albite in the Cˉ1 solid solution (γ _Ab ^Cˉ1 ) can be taken as 1.0. The possible importance of this model lies in its identification of the principal constraints on the mixing properties rather than in the actual values of γ _An ^Cˉ1 and γ _Ab ^Cˉ1 obtained. In particular it is recognised that γ _An ^Cˉ1 depends critically on ordering in anorthite as well as, at lower temperatures, any ordering in the Cˉ1 solid solution. A brief review of activity-composition data, from published experiments involving ranges of plagioclase compositions and from the combined heats of mixing plus Al-avoidance entropy model (Newton et al. 1980), reveals some inconsistencies. The values of γ _An ^Cˉ1 calculated using the approach of Newton et al. (1980), although consistent with Orville's (1972) ion exchange data, are slightly lower than values derived from experiments by Windom and Boettcher (1976) and Goldsmith (1982) or from ion-exchange experiments of Kotel'nikov et al. (1981). Based on the Cˉ1/Iˉ1 transformation model, values of γ _An ^Cˉ1 <1.0 are unlikely. Discrepancies between the experimental data sets are attributed to incomplete (non-equilibrium) Al/Si order attained during the experiments. It is suggested that the choice of activity coefficients remains somewhat subjective. The development of accurate mixing models would be greatly assisted by better thermodynamic data for ordering in pure anorthite and by more thorough characterisation of the state of order in plagioclase crystals used for phase equilibrium experiments.     

Raman microprobe study of synthetic diaplectic plagioclase feldspars

Raman microprobe spectra were made on three post shock, diaplectic plagioclase feldspars. Optical and X-ray diffraction studies indicated that feldspars maintained a partially or totally crystalline state after having passed through the mixed phase zone of Hugoniot response to shock waves (15–38 GPa). The appearance of uniquely glass-type spectra occurs at different shock pressures for each specimen according to its atomic structural arrangement, below 38 GPa for mosaic structured labradorite, near 40 GPa for anorthite and above 50 GPa for the highly ordered low albite. The diaplectic anorthite and labradorite glasses give spectra which indicate the presence of two glass types. Shifts in the band envelope frequencies compared to spectra of fused glass and statically pressure densified glass suggest that these glasses have specific structural arrangements. These differences suggest that the shock and fusion glass-forming processes are not exactly identical. The results from material shocked in the mixed phase region of Hugoniot response show that the phase transitions are effected at different pressures depending upon the feldspar structural type.     

Extinction-angle distribution of plagioclase feldspars in clastic sediments

An extinction-angle distribution was computed for randomly oriented grains of plagioclase feldspar with a known An-composition. Expressed by a histogram with eighteen columns, the results are shown by an 18×19 data array for nineteen types of low-temperature plagioclase. In a mixture of grains with various An-compositions, the extinction-angle distribution is given by a linear combination of the obtained data matrix with the frequencies of An-composition. If we have a sample containing randomly oriented grains of plagioclase feldspars, a frequency distribution of those with An-composition can be calculated after measuring their extinction angles. By taking into account the constraints of inequalities, the calculation is treated with the dual method of linear programming. The procedure provides for a simpler and more expedient method for determining the frequency distribution of An-composition. This method was tested by applying it to a model sample containing randomly oriented plagioclase grains with known An-composition frequencies, whose extinction-angle distribution was determined by the Monte Carlo method. An example of application of this method to Permian sandstones from central Japan is presented.     

Analysis of dislocations in some naturally deformed plagioclase feldspars

Dislocations in intermediate plagioclase feldspars, which were deformed under granulite facies conditions, have been analysed. The study reveals extensive ductile deformation by intracrystalline slip and by twinning. Six out of the seven possible Burgers vectors were identified: \(b = \left[ {001} \right],\tfrac{1}{2}\left[ {110} \right],\tfrac{1}{2}\left[ {1\bar 10} \right],\left[ {101} \right],\tfrac{1}{2}\left[ {112} \right]and\tfrac{1}{2}\left[ {1\bar 12} \right]\) . Most, perhaps all, dislocations are dissociated by up to 200 Å. The microstructure is dominated by [001] screw dislocations, most of which appear to be dissociated in (010). The dominant slip system appears to be (010) [001]. Large grain-to-grain variations in the density of free dislocations indicate that the plastic strain in individual grains depended upon the Schmid factor for (010) [001]. The microstructure suggests that the rate-controlling step for high-temperature creep of plagioclase is cross-slip of extended [001] screw dislocations. The rheological contrast between feldspar and quartz is partly due to a difference in stacking fault energy.     

Partitioning of Pb between volcanic glass and coexisting sanidine and plagioclase feldspars

Pb contents were determined by isotope dilution in separated glass, sanidine, and plagioclase from 18 rocks ranging in composition from basalt to rhyolite. These data indicate that Pb is partitioned into silicate melt relative to plagioclase, but is equally distributed between melt and sanidine. Plagioclase/glass distribution coefficients increase from 0.1 to 0.7 in going from basalt to rhyolite. This relationship suggests that the distribution coefficient is dependent upon bulk composition, temperature, or both. Sanidine/glass distribution coefficients are close to unity in rocks ranging in composition from quartz latite to rhyolite. The variation in Pb contents in a natural magma series from Craters of the Moon National Monument, Idaho, indicates that minerals (olivine, plagioclase, magnetite, apatite and clinopyroxene) fractionated from these magmas all have very low crystal/liquid distribution coefficients for Pb.     

The variation of tetrahedral bond lengths in sodic plagioclase feldspars

Multiple linear regression analysis has been applied to the geometric and chemical variables in sodic plagioclases in order to determine their relative effects on individual T-O bond lengths in the Al_1+xSi_3?xO₈ tetrahedral framework. Using data from crystal structure analyses of low and high albite, An₁₆ and An₂₈, and assuming that low albite is completely ordered, 1 $$\begin{gathered} {\text{T}} - {\text{O = 1}}{\text{.568}} + {\text{[(0}}{\text{.122) x (Al content of the T site)]}} \hfill \\ {\text{ }} - {\text{[(0}}{\text{.037) x (}}\Delta {\text{{\rm A}l}}_{{\text{br}}} )] + [0.063){\text{ x }}(\Sigma {\text{[}}q{\text{/(Na,Ca}} - {\text{O)}}^{\text{2}} ])] \hfill \\ {\text{ }} + {\text{[(0}}{\text{.029) x (}} - {\text{1/cosT}} - {\text{O}} - {\text{T)]}} \hfill \\ \end{gathered}$$ where the Al content of a particular tetrahedral (T) site can be estimated from empirically-derived determinative curves, where Δ Al_br is a linkage factor to account for the Al content of adjacent tetrahedral sites, where the formal charge on the (Na_1?xCa_x) atom is q=1+x, and where T-O-T is the inter-tetrahedral angle involving the T-O bond. For sodic plagioclases it is essential to know only the anorthite content and the 2Θ₁₃₁-2Θ_1¯31 spacing (CuK _α radiation) in order to determine the independent variables in this equation and thus to evaluate the individual T-O distances. The 64 individual T-O distances predicted for the four sodic plagioclases by this equation agree well with the observed T-O bond lengths (σ=0.004 Å; r=0.994), and the method has been used by way of example to rationalize the T-O bond lengths in analcime (cf. Ferraris, Jones and Yerkess, 1972).     

The replacement of plagioclase feldspars by albite: observations from hydrothermal experiments

Oligoclase and labradorite crystals have been experimentally replaced by albite in an aqueous sodium silicate solution at 600°C and 2 kbars. The replacement is pseudomorphic and is characterised by a sharp chemical interface which progresses through the feldspar while preserving the crystallographic orientation. Reaction rims of albite, up to 50 μm thick, can be readily achieved within 14 days. Re-equilibration of plagioclase in an ¹⁸O-enriched sodium- and silica-bearing solution results in oxygen isotope redistribution within the feldspar framework structure. The observed characteristics of the reaction products are similar to naturally albitised plagioclase and are indicative of an interface-coupled dissolution–reprecipitation mechanism. Chemical analyses demonstrate that the albitisation is accompanied by the mobilisation of major, minor and trace elements also including elements such as Al and Ti which are commonly regarded as immobile during hydrothermal alteration. The results contribute to developing our understanding of the close association between large-scale albitisation and secondary ore mineralisation which is common in nature.     

Sense of shear in high-temperature movement zones from the fabric asymmetry of plagioclase feldspars

A new method for determining the sense of shear in plagioclase-bearing tectonites from the (010) orientation of plagioclase feldspar is presented. The method is based on the asymmetry of the (010) plane with respect to the structural frame (foliation and lineation) and the dominant activity of the (010) slip plane in the high-temperature plasticity of plagioclase feldspar. Using examples from the Zabargad gneisses (Red Sea) the method is applied to plagioclases of An25–An45 and compared with other methods of shear-sense determination (quartz c-axis fabrics and microstructural criteria).     

Experimental delineation of the “e” ⇌ I\bar 1 and “e” ⇌ C\bar 1 transformations in intermediate plagioclase feldspars

Approximately 125 hydrothermal annealing experiments have been carried out in an attempt to bracket the stability fields of different ordered structures within the plagioclase feldspar solid solution. Natural crystals were used for the experiments and were subjected to temperatures of ～650°C to ～1,000°C for times of up to 370 days at \(P_{{\text{H}}_{\text{2}} {\text{O}}} \) =600 bars, or \(P_{{\text{H}}_{\text{2}} {\text{O}}} \) =1,200 bars. The structural states of both parent and product materials were characterised by electron diffraction, with special attention being paid to the nature of type e and type b reflections (at h+k=(2n+1), l=(2n+1) positions). Structural changes of the type C \(\bar 1\) → I \(\bar 1\) , C \(\bar 1\) → “e” structure, I \(\bar 1\) → “e” and “e” structure → I \(\bar 1\) have been followed. There are marked differences between the ordering behaviour of crystals with compositions on either side of the C \(\bar 1\) ? I \(\bar 1\) transition line. In the composition range ～ An₅₀ to ～ An₇₀ the e structure appears to have a true field of stability relative to I \(\bar 1\) ordering, and a transformation of the type I \(\bar 1\) ? e has been reversed. It is suggested that the e structure is the more stable ordered state at temperatures of ～ 800°C and below. For compositions more albite-rich than ～ An₅₀ the upper temperature limit for long range e ordering is lower than ～ 750°C, and there is no evidence for any I \(\bar 1\) ordering. The evidence for a true stability field for “e” plagioclase, which is also consistent with calorimetric data, necessitates reanalysis both of the ordering behaviour of plagioclase crystals in nature and of the equilibrium phase diagram for the albite-anorthite system. Igneous crystals with compositions of ～ An₆₅, for example, probably follow a sequence of structural states C \(\bar 1\) → I \(\bar 1\) → e during cooling. The peristerite, Bøggild and Huttenlocher miscibility gaps are clearly associated with breaks in the albite, e and I \(\bar 1\) ordering behaviour but their exact topologies will depend on the thermodynamic character of the order/disorder transformations.     

Transmission electron microscope study of the domain structures associated with the b-, c-, d-, e- and f-reflections in plagioclase feldspars

Ten specimens of plagioclase feldspars (An₁₀₀ to An₂₅) have been examined by transmission electron microscopy using dark field and direct lattice imaging. The specimens are classified into two groups: (i) those that have been quenched from the melt (synthetic An₁₀₀, An₉₅, An₆₆ and An₅₀) and (ii) those that have been cooled relatively slowly (An₉₄, An₇₇, An₇₅, An₅₂, An₃₂ and An₂₆). The observed contrast is interpreted in detail using the two-beam dynamical theory of electron diffraction contrast. The fault vectors of the antiphase boundaries (APBs) observed in An₁₀₀ and An₇₇ are determined. Direct resolution of the sublattice and superlattice in specimens of intermediate composition (An₇₅ to An₂₅) indicates that the superlattice is not a regular array of out-of-step faults and is probably due to a simple sinusoidal distortion of the structure. Only three specimens (An₁₀₀, An₇₇ and An₅₀) exhibited a single structural type. All the remaining specimens consisted of intergrowths of two structural types. The structural type present in any particular specimen depends upon its An-content and thermal history. The rate of transformation from any structural type to a lower-temperature structural type appears to decrease with decreasing An-content.     

On the constitution of the alkali feldspars

Summary The crystalline variants of the alkali feldspars differ from each other in the pattern of Al/Si disorder. The interrelations of the several variants are graphically shown in Figs. 1, 2, and 3. In high albite thekind of disorder is different from that in any other alkali feldspar; therefore the transition high albitemonalbite is diffusive; but the transition monalbiteanalbite is displacive. See Fig. 10.The phase diagram of the system NaAlSi₃O₈–KAlSi₃O₈ is presented in Fig. 4. The volume relations of the mixed crystals (Figs. 5 and 8) indicate that albite exhibiting monoclinic symmetry at room temperature has a defect lattice (defect monalbite) with vacant Na sites. A vacant cation site behaves as if a large cation (for example K) were present; therefore the defect monalbite behaves like a mixed crystal with some K replacing Na.

---

Zusammenfassung Die verschiedenen kristallinen Modifikationen der Alkalifeldspate unterscheiden sich bezüglich der Art und Weise ihrer Al/Si-Verteilung. Die Interrelationen sind in den Abb. 1, 2 und 3 dargestellt. Im Hochalbit ist die Al/Si-Verteilung andersartig als in irgendeinem anderen Alkalifeldspat. Die meisten Alkalifeldspate haben eine zwifache Unendlichkeit von Variationsmöglichkeiten der Al/Si-Verteilung (entsprechend der in Abb. 2 eingezeichneten Ebene), Hochalbit aber steht einzig da und weist eine dreifache Unendlichkeit von Möglichkeiten auf. Infolgedessen entspricht dem Übergang HochalbitMonalbit ein diffuser Umwandlungsmechanismus, dem Übergang MonalbitAnalbit aber ein displaciver (Abb. 10).Das Phasendiagramm des Systems NaAlSi₃O₈–KAlSi₃O₈ ist in Abb. 4 dargestellt. Die Volumverhältnisse der Mischkristalle (Abb. 5 und 8) deuten darauf hin, daß der bei gewöhnlicher Temperatur monokline Albit ein defektes Gitter hat (= defekter Monalbit mit teilweise unbesetzten Na-Punktlagen). Die Wirkung einer unbesetzten Kation-Punktlage im Gitter ist der Wirkung eines großen Kations (z. B. Kalium) gleich; hierdurch erklärt sich die Tatsache, daß der defekte Monalbit die Eigenschaften eines etwas K-haltigen Mischkristalls nachahmt.

---

---

With 10 Figures

---

Dedicated to ProfessorF. Machatschki on the occasion of his 70th birthday.     

Structure energies of the alkali feldspars

Structural energetics of the alkali feldspars have been studied using a “lattice” or structure energy model. Electrostatic energies, U _e,for 20 well-refined, non-intergrown alkali feldspars were calculated using Bertaut's (1952) summation procedure and average about ?13,400 kcal/mol; the repulsive energies of the alkali site in each structure (～15 kcal/mol) were calculated using repulsive parameters for K-O and Na-O interactions estimated from bulk modulus data for NaF and KF and the exponential form of the repulsive potential. Using a procedure in which the position of the alkali cation was varied while the oxygen cage was kept fixed, structure energy gradients for the alkali sites of high albite and a hypersolvus Ab₄₂Or₅₈ structure were computed. In both cases, a broad structure energy well, elongated approximately parallel to c and subparallel to the observed split Na positions, was found. In both structures there is a single energy minimum corresponding closely with the observed single alkali positions. Comparison of U _e values for the alkali feldspars with different K/Na ratios shows that intermediate compositions are predicted to be less “stable” than either endmember and that the potassic end-member is predicted to be less “stable” than the sodic one, assuming that all other factors contributiong to the free energies of each phase are approximately the same. Comparison of U _e values for the high albite and low sanidine structures with different Al/Si distributions and a fixed tetrahedral framework indicates that the ordered charge distributions are 63.0 and 54.8 kcal/mol, respectively, more “stable” than the disordered distributions. Smaller, more realistic energy differences were obtained by using U _evalues averaged from four separate calculations with a +3 charge on a different T site in each and with +4 charges on the other T sites. If, in addition, the charges on cations and oxygen are reduced to half their nominal formal charges, in agreement with Pauling's electroneutrality principle and the results of recent molecular orbital calculations on silicates, the predicted electrostatic energy differences are reduced to 3.6 and 1.6 kcal/mol, respectively. These calculations also indicate that the T₁O site in the high albite structure energetically favors Al and that the Al/Si distribution determines the Na position within the alkali site.     

Dissolution kinetics of plagioclase in the melt of the system diopside-albite-anorthite,and origin of dusty plagioclase in andesites

The textures and kinetics of reaction between plagioclase and melts have been investigated experimentally, and origin of dusty plagioclase in andesites has been discussed. In the experiments plagioclase of different compositions (An₉₆, An₆₁, An₅₄, An₂₃, and An₂₂) surrounded by glasses of six different compositions in the system diopside-albite-anorthite was heated at temperatures ranging from 1,200 to 1,410° C for 30 min to 88 h. Textures were closely related to temperature and chemical compositions. A crystal became smaller and rounded above the plagioclase liquidus temperature of the starting melt (glass) and remained its original euhedral shape below the liquidus. Whatever the temperature, the crystal-melt interface became rough and often more complicated (sieve-like texture composed of plagioclase-melt mixture in the scale of a few m was developed from the surface of the crystal inward; formation of mantled plagioclase) if the crystal is less calcic than the plagioclase in equilibrium with the surrounding melt, and the interface remained smooth if the crystal is more calcic than the equilibrium plagioclase. From these results the following two types of dissolution have been recognized; (1) a crystal simply dissolves in the melt which is undersaturated with respect to the phase (simple dissolution), and a crystal is partially dissolved to form mantled plagioclase by reaction between sodic plagioclase and calcic melt (partial dissolution). The amount of a crystal dissolved and reacted increased proportional to the square root of time. This suggests that these processes are controlled by diffusion, probably in the crystal.Mantled plagioclase produced in the experiments were very similar both texturally and chemically to some of the so-called resorbed plagioclase in igneous rocks. Chemical compositions and textures of plagioclase phenocrysts in island-arc andesites of magma mixing origin have been examined. Cores of clear and dusty plagioclase were clacic (about An₉₀) and sodic (about An₅₀), respectively. This result indicates that dusty plagioclases were formed by the partial melting due to reaction between sodic plagioclase already precipitated in a dacitic magma and a melt of intermediate composition in a mixed magma during the magma mixing.     

Optical studies of natural deformation microstructures in feldspars (gneiss and pegmatites from Occitania,southern France)

Natural deformation microstructures of feldspars have been investigated optically in augen gneisses and pegmatites from southern France that exhibit microstructures due to polyphase deformation. This deformation was produced in both cases under a pressure of about 2 Kb and a temperature decreasing from about 550° to 200–300°C.In microcline, recrystallization, tension gashes and shear planes are the most important microstructures observed. In plagioclase, shear planes and kink-bands prevail. The development of these structures is controlled primarily by three parameters: temperature, influence of the surrounding material and amount of bulk strain. The orientation of the planes of anisotropy of the minerals with respect to the compression axis is also important. Some perthitic patterns are ascribed to deformation.     

Superplastic flow and changes in crystal chemistry of feldspars

Comparison of observed microstructures and deduced physical parameters of natural deformations with those encountered in superplastic metals permits identification of superplastic behaviour in rocks. From microstructural evidence, superplastic flow of feldspar is inferred to have occurred in the deformation of a granite from Miéville, Aiguilles-Rouges massif, Switzerland. The original igneous orthoclase and oligoclase phenocrysts have both recrystallised to fine-grained albite which has deformed by superplastic flow. This change in crystal chemistry was associated with a large negative change in free energy, enabling recrystallised grains of a very small size to be stable under the prevailing conditions, thereby producing the pre-requisite conditions for superplastic behaviour.     

On the mechanisms of cation diffusion processes in ternary feldspars

Cation diffusion processes have been studied in single crystals of intermediate plagioclase and albite composition by tracer-diffusion experiments and optical absorption spectroscopy. Tracer-diffusivities were determined by the residual activity method, using the radioactive isotopes ²²Na, ⁴⁵Ca and ⁵⁹Fe. In most cases, diffusion experiments were performed at 1 bar, at controlled oxygen activity and at temperatures between 750 and 1300°C. The obtained Na-diffusivities for plagioclases were much smaller then previously determined for albite. This indicates a strong composition dependence of Na-diffusion. In contrast, Ca-diffusivity in albite does not differ very much from that in intermediate plagioclases. The relative diffusivities determined for plagioclase of composition An62 at 1200° C (CO/CO₂ =50∶50) were D _infNa ^sup* ∶D _infFe ^sup* ∶D _infCa ^sup* =5000∶10∶1. Despite the an isotropic structure of feldspars, no difference was found for Na-and Ca-diffusion normal to (001) and normal to (010). Water pressure of 2 kbar has no influence on the Na-diffusivity. In contrast to the Ca-diffusion, a dependence on oxyggen activity was found for Na-and Fe-diffusion. Fe-diffusivity increases with decreasing oxygen activity. This can be correlated to changes in oxidation state of iron dissolved in the plagioclases. Optical absorption spectroscopy shows that iron is oxidized in the plagioclases by annealing in air. This effect can be reversed by annealing at reducing conditions. A model is proposed to explain the oxidation of iron by a chemical diffusion process in which A-vacancies are formed by out-diffusion of Na⁺. Preannealing of samples in air gives a temperature independent decrease of Na-diffusivity by a factor of about 2.5. This effect is explained with help of a simple disorder model for A-cations in ternary feldspars. It is concluded that Na⁺ diffuses via interstitials and that the A-vacancy concentration in the plagioclases is controlled extrinsically, probably by dissolved SiO₂.