Damage and Changes in Mechanical Properties of a Gabbro Thermally Loaded up to 1,000°C

Thermal loading of rocks at high temperatures induces changes in their mechanical properties. In this study, a hard gabbro was tested in the laboratory. Specimens were slowly heated to a maximum temperature of 1,000°C. Subsequent to the thermal loading, specimens were subjected to uniaxial compression. A drastic decrease of both unconfined compressive strength and elastic moduli was observed. The thermal damage of the rock was also highlighted by measuring elastic wave velocities and by monitoring acoustic emissions during testing. The micromechanisms of rock degradation were investigated by analysis of thin sections after each stage of thermal loading. It was found that there is a critical temperature above which drastic changes in mechanical properties occur. Indeed, below a temperature of 600°C, microcracks start developing due to a difference in the thermal expansion coefficients of the crystals. At higher temperatures (above 600°C), oxidation of Fe²⁺ and Mg²⁺, as well as bursting of fluid inclusions, are the principal causes of damage. Such mechanical degradation may have dramatic consequences for many geoengineering structures.     

Experimental investigation of the strength and failure behavior of layered sandstone under uniaxial compression and Brazilian testing

As a typical inherently anisotropic rock, layered sandstones can differ from each other in several aspects, including grain size, type of material, type of cementation, and degree of compaction. An experimental study is essential to obtain and convictive evidence to characterize the mechanical behavior of such rock. In this paper, the mechanical behavior of a layered sandstone from Xuzhou, China, is investigated under uniaxial compression and Brazilian test conditions. The loading tests are conducted on 7 sets of bedding inclinations, which are defined as the angle between the bedding plane and horizontal direction. The uniaxial compression strength (UCS) and elastic modulus values show an undulatory variation when the bedding inclination increases. The overall trend of the UCS and elastic modulus values with bedding inclination is decreasing. The BTS value decreases with respect to the bedding inclination and the overall trend of it is approximating a linear variation. The 3D digital high-speed camera images reveal that the failure and fracture of a specimen are related to the surface deformation. Layered sandstone tested under uniaxial compression does not show a typical failure mode, although shear slip along the bedding plane occurs at high bedding inclinations. Strain gauge readings during the Brazilian tests indicate that the normal stress on the bedding plane transforms from compression to tension as the bedding inclination increases. The stress parallel to the bedding plane in a rock material transforms from tension to compression and agrees well with the fracture patterns; “central fractures” occur at bedding inclinations of 0°–75°, “layer activation” occurs at high bedding inclinations of 75°–90°, and a combination of the two occurs at 75°.     

Observation of Post-failure Kaiser Effect in a Plastic Rock

—?Cyclic uniaxial and triaxial tests of rock salt samples were carried out in pre- and post-failure regimes. In all tests, a distinct Kaiser effect was observed in the post-failure region. The slope of the curve “cumulative AE counts versus axial strain” increased dramatically (several times to an order) when the strain attained its peak value achieved previously. Neither preliminary hydrostatic compression nor preliminary uniaxial cyclic pre-failure loading influence the post-failure Kaiser effect. The results for the plastic rock are opposite to the familiar absence of the Kaiser effect in brittle rocks at stress values approaching the ultimate strength. The results obtained during laboratory tests of a plastic rock are of importance for the proof of a connection between the Kaiser effect and foreshocks preceding earthquakes.     

Variations of Strength,Resistivity and Thermal Parameters of Clay after High Temperature Treatment

This paper reports the variations of strength, resistivity and thermal parameters of clay after high-temperature heating. Experiments were carried out to test the physical properties of clay heated at temperatures ranging from room temperature to 800°C in a furnace. The experiment results show that below 400°C the uniaxial compressive strength and resistivity change very little. However, above 400°C, both increase rapidly. At a temperature under 400°C, the thermal conductivity and specific heat capacity decrease significantly. The thermogravimetric analysis (TG) and differential scanning calorimeter (DSC) test indicate that a series of changes occur in kaolinite at temperatures from 400 to 600°C, which is considered the primary cause of the variation of physical and mechanical properties of clay under high temperatures.     

Measurements of thermal conductivity and specific heat in volcanogenic rocks

A laboratory installation has been developed together with a technique for determining thermo-physical properties (thermal conductivity and specific heat) in cylindrical rock specimens. The technique is based on iTOUGH2-EOS3 inversion modeling using temperature measurements inside specimens as a result of their short-term heating and subsequent return to the initial temperature. We estimated the thermal conductivity and specific heat for a collection of volcanogenic petrotypes that reflect the rocks that compose the Rogozhnikovskii volcanogenic oil reservoir (29 specimens). The average thermal conductivity of the dry rocks is 1.47 W/m °C and the average specific heat is 754 kJ/kg °C; the reproducibility of this estimation is 2.2% for thermal conductivity and 0.7% for specific heat.     

Eclogites from the Chinese continental scientific drilling borehole, their petrology and different P-T evolutions

Abstract Four phengite‐bearing eclogites, taken from different depths of the Chinese continental scientific drilling (CCSD) borehole in the Sulu ultrahigh pressure terrane, eastern China, were studied with the electron microprobe. The compositional zonations of garnet and omphacite are moderate, whereas phengite compositions generally vary significantly in a single sample from core to rim by decrease of the Si content. Various geothermobarometric methods were applied to constrain the P‐T conditions of these eclogites on the basis of the compositional variability of the above minerals. The constrained P‐T path for sample B218 is characterized by pressure decrease from ca 3.0 GPa (ca 600°C) to 1.3 GPa (ca 550°C). Eclogite B310 yielded P‐T conditions of 3.0 GPa and 750°C. The path for eclogite B1008 starts at about 650°C and 3.6–3.9 GPa (stage I) followed by a pressure decrease to 2.8–3.0 GPa and a significant temperature rise (stages II and IIIa, 750–810°C). Afterwards, this rock cooled down to 620–660°C at still high pressures (2.5–2.7 GPa, stage IIIb). Retrograde conditions were about 670°C and 1.3 GPa (stage IV). Eclogite B1039 yielded a P‐T path starting at ca 600°C and 3.3–3.9 GPa (stage I). A pressure decrease to about 3.0 GPa (stage II, 590–610°C) and then a moderate isobaric temperature increase to ca 630°C (stage III) followed. Stage IV is characterized by temperatures of 650°C at pressures close to 1.3 GPa. During and after this stage (hydrous) fluids partially rich in potassium penetrated the rocks causing minor changes. Relatively high oxygen fugacities led to andradite and magnetite among the newly formed minerals. We think that the above findings can be best explained by mass flow in a subduction channel. Thus, we conclude that the assembly of UHP rocks of the CCSD site, eclogites, quartzofeldspathic rocks, and peridotites, cannot represent a crustal section that was already coherent at UHP conditions as it is the common belief currently. The coherency was attained after significant exhumation of these UHP rocks.     

Numerical and Experimental Study on Progressive Failure of Marble

This study presents a framework for numerical simulations based upon micromechanical parameters in modeling progressive failures of heterogeneous rock specimens under compression. In our numerical simulations, a Weibull distribution of the strength and elastic properties of the finite elements is assumed, and the associated Weibull parameters are estimated in terms of microstructural properties, such as crack size distribution and grain size, through microscopic observations of microcracks. The main uncertainty in this procedure lies on the fact that various ways can be used to formulate a ``microcrack size distribution' in relating to the Weibull parameters. As one possible choice, the present study uses the number of counted cracks per unit scanned volume per grain size to formulate the crack distributions. Finally, as a tool, the Rock Failure Process Analysis code (RFPA^2D) is adopted for simulating the progressive failure and microseismicity of heterogeneous rocks by using an elastic-damage finite-element approach. To verify our framework, compression tests on marble specimens are conducted, and the measured acoustic emissions (AE) are compared with those predicted by the numerical simulations. The mode of failure, compressive strength and AE pattern of our simulations basically agree with experimental observations.     

Application of low-field hysteresis and memory-phenomenon studies in determining the stability of magnetization of Deccan Trap basalts from Mount Girnar and Mount Pavagarh regions,India

Two techniques, namely alternating-field demagnetization and thermal demagnetization, are widely being used for determining the stability of magnetization of a rock specimen. Recently a faster and simpler technique known as low-field hysteresis loop and memory-phenomenon test has been developed for determining the stability of magnetization directions of igneous rocks. In this paper the results of this new technique, after applying it to about 1000 specimens obtained from 250 oriented rock samples collected from 42 sites of Deccan Trap basalts from Mount Girnar (21°30′N; 70°30′E) and Mount Pavagarh (22°30′N; 73°30′E), India, are presented.The agreement between the mean natural remanent magnetization directions determined by this procedure and those computed after alternating-field demagnetization has been found to be very good. All the specimens from Mount Girnar and 94% of Mount Pavagarh specimens showed a stable line without any memory in a field ≈ 10 Oe. This indicates that the rocks from these two localities are highly stable and are most suitable ones for the determination of a precise palaeomagnetic direction.     

Influence of Outcrop Scale Fractures on the Effective Stiffness of Fault Damage Zone Rocks

We combine detailed mapping and microstructural analyses of small fault zones in granodiorite with numerical mechanical models to estimate the effect of mesoscopic (outcrop-scale) damage zone fractures on the effective stiffness of the fault zone rocks. The Bear Creek fault zones were active at depths between 4 and 15 km and localize mesoscopic off-fault damage into tabular zones between two subparallel boundary faults, producing a fracture-induced material contrast across the boundary faults with softer rocks between the boundary faults and intact granodiorite outside the boundary faults. Using digitized fault zone fracture maps as the modeled fault geometries, we conduct nonlinear uniaxial compression tests using a novel finite-element method code as the experimental “laboratory” apparatus. Map measurements show that the fault zones have high nondimensional facture densities (>1), and damage zone fractures anastamose and intersect, making existing analytical effective medium models inadequate for estimation of the effective elastic properties. Numerical experiments show that the damage zone is strongly anisotropic and the bulk response of the fault zone is strain-weakening. Normal strains as small as 2% can induce a reduction of the overall stiffness of up to 75%. Fracture-induced effective stiffness changes are large enough to locally be greater than intact modulus changes across the fault due to juxtaposition of rocks of different lithologies; therefore mesoscopic fracturing is as important as rock type when considering material or bimaterial effects on earthquake mechanics. These results have important implications for earthquake rupture mechanics models, because mesoscopic damage zone fractures can cause a material contrast across the faults as large as any lithology-based material contrast at seismogenic depths, and the effective moduli can be highly variable during a single rupture event.     

Effects of pore fluids on quasi-static shear modulus caused by pore-scale interfacial phenomena

It is evident from the laboratory experiments that shear moduli of different porous isotropic rocks may show softening behaviour upon saturation. The shear softening means that the shear modulus of dry samples is higher than of saturated samples. Shear softening was observed both at low (seismic) frequencies and high (ultrasonic) frequencies. Shear softening is stronger at seismic frequencies than at ultrasonic frequencies, where the softening is compensated by hardening due to unrelaxed squirt flow. It contradicts to Gassmann's theory suggesting that the relaxed shear modulus of isotropic rock should not depend upon fluid saturation, provided that no chemical reaction between the solid frame and the pore fluid. Several researchers demonstrated that the shear softening effect is reversible during re-saturation of rock samples, suggesting no permanent chemical reaction between the solid frame and the pore fluid. Therefore, it is extremely difficult to explain this fluid–rock interaction mechanism theoretically, because it does not contradict to the assumptions of Gassmann's theory, but contradicts to its conclusions. We argue that the observed shear softening of partially saturated rocks by different pore fluids is related to pore-scale interfacial phenomena effects, typically neglected by the rock physics models. These interface phenomena effects are dependent on surface tension between immiscible fluids, rock wettability, aperture distribution of microcracks, compressibility of microcracks, porosity of microcracks, elastic properties of rock mineral, fluid saturation, effective stress and wave amplitude. Derived equations allow to estimate effects of pore fluids and saturation on the shear modulus and mechanical strength of rocks.     

Fracture toughness and subcritical crack growth during high-temperature tensile deformation of Westerly granite and Black gabbro

The double torsion testing method has been used to determine catastrophic and subcritical crack propagation parameters for pre-cracked specimens of Westerly granite and Black gabbro under a number of environmental conditions.The critical stress intensity factor for catastrophic crack propagation (fracture toughness) of granite and gabbro has been measured at temperatures from 20 to 400°C, in a vacuum. At 20°C, the fracture toughness of Westerly granite was $\text{1.79 ± 0.02}\text{MPa}\text{·}\text{m}{}^{\mathrm{<mtext>1</mtext><mtext>2</mtext>}}$, and for two blocks of Black gabbro it was $\text{3.03 ± 0.08}\text{MPa}\text{·}\text{m}{}^{\mathrm{<mtext>1</mtext><mtext>2</mtext>}}$ and $\text{2.71 ± 0.15}\text{MPa}\text{·}\text{m}{}^{\mathrm{<mtext>1</mtext><mtext>2</mtext>}}$, respectively. These values are very close to those reported by other investigators for tests conducted in air of ambient humidity at room temperature. For both rocks, fracture toughness at first increased slightly, and then decreased steadily on raising the temperature above ambient conditions. This behaviour is explained in terms of the density and distribution of thermally induced microcracks, as determined by quantitative optical microscopy.Subcritical crack growth behaviour has been studied at temperatures up to 300°C, and under water vapour at pressures of 0.6 to 15 kPa. Both the load relaxation and incremental constant displacement rate forms of the double torsion testing method were utilised to generate stress intensity factor/crack velocity diagrams. Crack growth was measured over the velocity range 5 × 10^?3 to 10^?7 m · s^?1. Increasing both temperature and water vapour pressure resulted in substantially higher crack growth rates. The overall effect of raising the temperature over the range studied here (20–300°C) was to increase the crack growth rate in granite and gabbro by ～5 and 7 orders of magnitude, respectively, at constant stress intensity factor and vapour pressure of water. For both rocks, the slopes of stress intensity factor/crack velocity curves were sensitive to changes in both temperature and water vapour pressure at low values of the latter parameter. Slopes fell substantially on raising the water vapour pressure, but were relatively insensitive to changes in temperature at these higher pressures. No subcritical crack growth limit was encountered.Estimates of the uncertainty in our experimental data are given. From the results of multiple load relaxation experiments on Westerly granite specimens, we estimate the uncertainty in position of stress intensity factor/crack velocity curves along the stress intensity axis to be c. 10% of the fracture toughness, and the uncertainty in slope of such curves to be c. 12%.Problems associated with the extrapolation of our experimental data to regions of higher effective confining pressure in the Earth's crust are discussed.     

Relation of fracture resistance to fabric for granitic rocks

Double-torsion specimens of two granitic rocks were prepared in several directions with reference to microcracks fabric. Even for the same rock and at the same stress levels, the observed crack velocities in two granitic rocks were dependent on both the propagation direction and the opening direction. The maximum difference by several orders of magnitude was found for both rocks. The highest crack velocity was observed when the subcritical crack was parallel to most of the preexisting cracks. The maximum critical stress intensity factor was about twice as high as the minimum one in different directions. An analysis for a thin plate having anisotropic elasticity under torsional load showed that the observed difference in the crack velocity and the critical stress intensity factor was not an error due to conventional equations derived on the assumption of isotropic elasticity but the true material's property. As the preferred orientation of microcracks has been pointed out for many granitic rocks, we can conclude that the anisotropic nature of the fracture resistance of the two granitic rocks used in this study was not exceptional. A region of a transport-limited velocity was not found for rocks, even at the velocity of 10^–2 m/s, that is almost equal to the theoretical limit of the stress corrosion cracking.     

Author index 1984 (volumes 67–71)

Magnetic properties of samples from Bell Island sedimentary rocks have been studied. X-ray analysis indicates that the main magnetic mineral is hematite in all samples. The other iron-bearing minerals identified are siderite and chamosite. Microscope observations of thin sections suggest that the rocks consist of oolitic hematite in a matrix of siderite or calcite. The intensity of natural remanent magnetization (NRM) varies in the range of (0.03–0.4 A m^?1), depending on the percentage of hematite. The thermal demagnetization curves of NRM show in some cases a sharp increase in magnetization at temperatures in the range 500–600°C. The peaks that occur in these demagnetization curves are due to a chemical change of siderite during repeated laboratory heating. X-ray analysis confirmed that the newly formed material is magnetite. Since the original NRM has been masked by the new intergrown material, this would result in a serious error in the determination of paleomagnetic pole positions. The samples showing this behaviour were not considered for paleomagnetic study. The samples containing oolitic hematite in a calcite matrix exhibit very high stability of NRM, including directional stability until almost 670°C. For these samples, a virtual pole position based on N = 6 samples (32 specimens) demagnetized to 665°C is 34°N, 114°E, not far from published Ordovician poles for the North American craton.     

Spinifex-like textured metaperidotites from the Higo Metamorphic Rocks,Japan, a possible high-pressure dehydration product of antigorite serpentinite

Spinifex-like textured metaperidotites from the Higo Metamorphic Rocks (HMR), west-central Kyushu, Japan, may be formed by high-pressure dehydration of antigorite, and may indicate deep subduction of serpentinite reaching a pressure–temperature condition of 1.6 GPa and 740–750 °C. Three rock types have been identified based on mineral assemblage and rock texture: Type I (L) consisting of medium-grained (1–5 cm long) olivine + enstatite + chromite ±tremolite with secondary talc and anthophyllite that occurs in low-grade metamorphic rocks of the biotite zone, Type I (H) of coarse-grained (up to 10 cm long) olivine + enstatite (with clinoenstatite lamella) + chromite ±tremolite with secondary talc that occurs in high-grade metamorphic rocks of the garnet-cordierite zone, and Type II composed of Al-spinel + chlorite + olivine + apatite + ilmenite with minor sodic gedrite in the garnet-cordierite zone together with Type I (H). Olivines in all rock types are mostly serpentinized during exhumation. The chromite-olivine thermometer gives 560–690 °C for Type I (L) rocks, and the spinel-olivine thermometer gives 610–740 °C for Type II rocks. The peak metamorphic pressure will be higher than 1.6 GPa based on the location of the experimentally determined invariant point (P = 1.6 GPa and T = 670 °C) of antigorite + forsterite + enstatite + talc + H₂O. This estimate is consistent with the occurrence of chlorite in Type II rocks, which is stable up to 890 °C at 2.0 GPa. The spinifex-like textured metaperidotites occur as small bodies in the low P/T type gneisses, implying tectonic juxtaposition of them probably during exhumation of the HMR. Recent findings of medium pressure (0.9–1.2 GPa) granulites and gneisses from the HMR may indicate that the HMR has a deep root into the wedge mantle from which the spinifex-like textured metaperidotites have derived.     

Modeling past and future alpine permafrost distribution in the Colorado Front Range

Rock glaciers, a feature associated with at least discontinuous permafrost, provide important topoclimatic information. Active and inactive rock glaciers can be used to model current permafrost distribution. Relict rock glacier locations provide paleoclimatic information to infer past conditions. Future warmer climates could cause permafrost zones to shrink and initiate slope instability hazards such as debris flows or rockslides, thus modeling change remains imperative. This research examines potential past and future permafrost distribution in the Colorado Front Range by calibrating an existing permafrost model using a standard adiabatic rate for mountains (0·5 °C per 100 m) for a 4 °C range of cooler and warmer temperatures. According to the model, permafrost currently covers about 12 per cent (326·1 km²) of the entire study area (2721·5 km²). In a 4 °C cooler climate 73·7 per cent (2004·4 km²) of the study area could be covered by permafrost, whereas in a 4°C warmer climate almost no permafrost would be found. Permafrost would be reduced severely by 93·9 per cent (a loss of 306·2 km²) in a 2·0 °C warmer climate; however, permafrost will likely respond slowly to change. Relict rock glacier distribution indicates that mean annual air temperature (MAAT) was once at least some 3·0 to 4·0 °C cooler during the Pleistocene, with permafrost extending some 600–700 m lower than today. The model is effective at identifying temperature sensitive areas for future monitoring; however, other feedback mechanisms such as precipitation are neglected. Copyright © 2005 John Wiley & Sons, Ltd.     

Loading Rate Dependence of Tensile Strength Anisotropy of Barre Granite

Granitic rocks usually exhibit strongly anisotropy due to pre-existing microcracks induced by long-term geological loadings. The understanding of the rock anisotropy in mechanical properties is critical to a variety of rock engineering applications. In this paper, Brazilian tests are conducted statically with a material testing machine and dynamically with a split Hopkinson pressure bar system to measure both static and dynamic tensile strength of Barre granite. To understand the anisotropy in tensile strength, samples are cored and labelled using the three principle directions of Barre granite to form six sample groups. For dynamic tests, a pulse shaping technique is used to achieve dynamic equilibrium in the samples during the dynamic test. The finite element method is then implemented to formulate equations that relate the failure load to the material tensile strength by employing an orthotropic elastic material model. For samples in the same orientation group, the tensile strength shows clear loading rate dependence. The tensile strengths also exhibit clear anisotropy under static loading while the anisotropy diminishes as the loading rate increases, which may be due to the interaction of pre-existing microcracks.     

Brittle crack growth in rocks

Fracture phenomena in rocks are associated with mainly mode I crack growth, sometimes superposed by shear or torsion. The present paper contributes to a fracture mechanics analysis of mode I and mixed mode crack propagation, by presenting reliable fracture toughness data for some rocks which include the effect of induced crack propagation rate, and the influence of effective pressure, and by numerical calculations on fracture propagation in layered rock formations. Empirical relations between fracture toughness,K _Ic' and induced crack opening displacement rate, as well as effective pressure, are given. The observedK _Ic pressure relation supports a theoretical model which takes into account the existence of microcracks in the crack tip region. Finite element calculations of fracture propagation in layered rock formations demonstrate the important effect of mixed mode crack growth. The numerical approach is particularly applied to single crack growth in hydraulic fracturing and in three point bending tests on layered single edge crack specimens.     

The influence of solar‐induced thermal stresses on the mechanical weathering of rocks in humid mid‐latitudes

The role of solar‐induced thermal stresses in the mechanical breakdown of rock in humid‐temperate climates has remained relatively unexplored. In contrast, numerous studies have demonstrated that cracks in rocks found in more arid mid‐latitude locations exhibit preferred northeast orientations that are interpreted to be a consequence of insolation‐related cracking. Here we hypothesize that similar insolation‐related mechanisms may be efficacious in humid temperate climates, possibly in conjunction with other mechanical weathering processes. To test this hypothesis, we collected rock and crack data from a total of 310 rocks at a forested field site in North Carolina (99 rocks, 266 cracks) and at forested and unforested field sites in Pennsylvania (211 rocks, 664 cracks) in the eastern United States. We find that overall, measured cracks exhibit statistically preferred strike orientations (47° ± 16), as well as dip angles (52° ± 24°), that are similar in most respects to comparable datasets from mid‐latitude deserts. There is less variance in strike orientations for larger cracks suggesting that cracks with certain orientations are preferentially propagated through time. We propose that diurnally repeating geometries of solar‐related stresses result in propagation of those cracks whose orientations are favorably oriented with respect to those stresses. We hypothesize that the result is an oriented rock heterogeneity that acts as a zone of weakness much like bedding or foliation that can, in turn, be exploited by other weathering processes. Observed crack orientations vary somewhat by location, consistent with this hypothesis given the different latitude and solar exposure of the field sites. Crack densities vary between field sites and are generally higher on north‐facing boulder‐faces and in forested sites, suggesting that moisture‐availability also plays a role in dictating cracking rates. These data provide evidence that solar‐induced thermal stresses facilitate mechanical weathering in environments where other processes are also likely at play. Copyright © 2015 John Wiley & Sons, Ltd.     

Cation distribution in low-calcium pyroxenes: Dependence on temperature and calcium content and the thermal history of lunar and terrestrial pigeonites

Four pyroxenes with compositions En₄₈Fs₄₈Wo₄, En_47·5Fs_47·5Wo₅, En₄₅Fs₄₅Wo₁₀ and En₄₀Fs₄₀Wo₂₀, synthesized at 1200°C at atmospheric pressure, were heat-treated at 500, 600, 700, and 800°C for various lengths of time. These pyroxenes are variously ordered with respect to Fe²⁺ and Mg²⁺ without unmixing. The Fe²⁺-Mg²⁺ distribution over the two nonequivalent sites M1 and M2, determined through Mössbauer spectroscopy, is found to be a function of both temperature and concentration of Ca²⁺ at the M2 site. The preference of Fe²⁺ for the M2 site increases with decreasing temperature and increasing Ca²⁺. These data can be used to determine cation equilibration temperatures of lunar and terrestrial pigeonites. The lunar pigeonites usually indicate equilibration temperatures of 700–860°C, except the pigeonite from rock 14053, which may have been subjected to shock heating due to meteoritic impact.     

The effect of cracks on the thermal expansion of rocks

Thermal expansion during the first heating cycle at atmospheric pressure was measured in several directions in seven igneous rocks between 25° and 400°C at slow heating rates. The coefficient of thermal expansion measured under these conditions increases more rapidly as temperature is increased than the average thermal expansion coefficient of the constituent minerals. The “extra” expansion is attributed to the formation of cracks by differential expansion of mineral grains. The presence of such cracks in the rocks during the cooling part of the cycle and during any subsequent heating and cooling cycles will result in a substantial decrease in the coefficient of thermal expansion as compared to that measured during the first heating cycles. The effect of cracks initially present in a rock was studied by measuring the full tensor of the coefficient of thermal expansion on two rocks with anisotropic crack distributions. In these two rocks the coefficient of thermal expansion is least in the direction perpendicular to the plane of greatest crack concentration. The implication of our data is that thermal expansion depends greatly on the fracture state of the rock. Both the fractures in the rock and the boundary conditions on the rock are significant for the interpretation of thermal expansion measurements and for their application to other problems.