Mass extinctions are important events for the evolution of life on Earth but often the mechanisms behind them are poorly understood. The Eocene - Oligocene Transition (EOT) had a profound and lasting influence on global climate and, though not one of the “big 5” mass extinction events, widespread extinctions in many fossil groups have been recognised. Larger benthic foraminifera (LBF), one of the most conspicuous and widespread fossil groups during the Eocene, are known to have experienced a rapid global overturning during the EOT, including the extinction of long-ranging families such as the Discocyclinidae, Orbitoclypeidae, Pellatispiridae and a number of species in the Nummulitidae. However, detailed records through the transition are rare, and few complete sections are known; the timing and causes of extinctions therefore remain uncertain. Extensive field samples from the southern Tanzanian coastal region along with Tanzania Drilling Project (TDP) samples are used to give an overview of Eocene to Miocene LBF and to produce high resolution stratigraphic records of LBF events across the EOT. Two further important EOT sites, Fuente Caldera, Spain and Melinau Gorge, Sarawak, both with unresolved questions relating to LBF at the EOT were studied for comparison. The field samples combined with the TDP data cover an area of approximately 200 km from Kilwa to the Mozambique border. Initial comparisons with Tethyan and Indo-Pacific faunas show perhaps an intermediate assemblage in the Eocene and some similarities with the Indo-Pacific in the Miocene. However in both epochs there are features distinctive of an East African or western Indo-Pacific fauna. Additionally, this work identifies many previously unreported localities and provides a solid basis for future work. Three of the TDP sites span the EOT and contain abundant LBF. Extensive calcareous micro-, nannofossil and stable isotope studies of these cores allow the LBF stratigraphy to be tied to global stratigraphy. These records show that the LBF extinction event occurs close to the Eocene/Oligocene boundary (EOB), as defined by the extinction of the planktonic foraminiferal Family Hantkeninidae, rather than at the prominent oxygen isotope excursion in the early Oligocene that signifies maximum ice growth and global sea-level fall. New bulk isotope data from the Melinau Limestone of Sarawak further support this conclusion. In Fuente Caldera, where the extinction level was previously reported to be within the Oligocene, extensive reworking means that an EOB extinction is also likely at this site. This correlation raises new questions about the cause of the extinctions and has important implications for global larger benthic foraminiferal stratigraphy.

The Utsira Sand is a major North Sea Saline aquifer. It is currently being utilised as a CO2 sequestration reservoir for the Sleipner project, and its large estimated storage capacity and proximity to suitable infrastructure ensures that it is an attractive prospect for further storage projects. As a forerunning CCS project it is the subject for scrutiny from environmental, industrial and governmental regulatory bodies, with the key concern being the confinement of the CO2 within the reservoir. Regulations are in place to ensure that operators monitor closely the migration of the CO2 plume within storage reservoirs and that they can demonstrate that the plume is behaving as modelled. Any unexpected behaviour of the plume requires escalation of the monitoring program and leakage would involve financial penalties in addition to remediation costs. Diligent site selection and a complete and robust understanding of the reservoir, adjacent units, and the overburden is therefore critical pre-injection. Geological features which have become apparent following CO2 injection into the Utsira Sand would suggest that the injection site was not completely understood before injection commenced. This study has contributed to a better understanding of the storage reservoir, specifically in regards to the deformed base of the Utsira Sand, which has been shown to be the result of sand remobilisation from depth in the form of sand intrusions. These intrusions, hosted within low permeability shales, represent permeable migration pathways from the deeper subsurface to the base of the Utsira Sand. Furthermore, their post-emplacement compaction has facilitated subsidence of the overlying strata. This movement, shown to affect strata up to the top of the Utsira Sand, has formed a ring of faults which projects obliquely upwards from the base of the Utsira Sand to the top of the storage reservoir. Amplitude anomalies within the overburden suggest that these zones of increased permeability have been exploited as paleo-gas migration pathways. Other amplitude anomalies within the succession overlying the Utsira Sand are also recognised. These include: high amplitude anomalies interpreted as gas accumulations within the Lower Seal, vertically focused ‘wipe out zones’ and zones of disrupted reflections interpreted as gas chimneys. The gas chimneys in particular, shown to extend to the seafloor, pose a significant leakage risk to sequestered CO2. On the basis of these observations, a plumbing system, from the Mid-Miocene to the seafloor, is presented. Potential CO2 leakage scenarios are proposed and discussed in respect to the storage ‘site’ and storage ‘complex’. It is suggested that potential future operators within this area should be aware of the features observed and appreciate the associated risk to stored CO2. It is also concluded that the viability of carbon sequestration technology as a climate change mitigation option requires that it represents an attractive proposition for those that operate the technology. Therefore it is considered that the regulation of carbon sequestration requires a finely measured balance between careful regulation and appropriate penalties for poor practice and flexibility in the interpretation of a ‘storage complex’ and ‘leakage.

This research project used 3D seismic data located in deep water fold and thrust belt in the Levant Basin eastern Mediterranean, to investigate the nature and kinematics of compartmentalized thrust related folds. The principal aim is to better understand thrust related fold development and interactions in compressional settings. The fold and thrust belt in the Levant Bain is mainly comprised of overlapping thrust faults of similar and opposing dips segmented or bounded by conjugate sets of strike slip faults. Detailed interpretation and analysis of the 3D geometry of the structures revealed that thrust faulting is an early process in the development of the thrust and fold pair, thrust interact with each other, and strike slip faults along strike. A preliminary end member interaction of thrust faults and strike slip faults is proposed based on observation of their bounding or segmenting pattern. The concept of fault interaction was mainly developed from the investigation of the propagation of thrust fault compartmentalised by strike slip faults. This involves a combination of kinematic analysis which includes fault displacement and shortening profiles, and the patterns of syn kinematic sediments above fold limb. Kinematic data suggests that strike slip faults are acting as barriers to thrust fault propagation. Similar barrier to fault propagation are observed between overlapping thrust faults within a single fold formed by the linkage of smaller thrust folds. The results showed that the faults are restricted as they link and transfer displacement. In addition to the propagation of thrust faults, vertical distribution of fault displacement suggests that they ramp up from detachment, this agrees with the classical models of thrust propagation folds.

Fluid venting structures are used to evaluate fluid migration in the subsurface and vertical changes in their morphology reflect variations in the intensity of fluid leakage through time. This research uses high-resolution 3D seismic data from the Lower Congo Basin offshore Angola to analyse complex assemblages of vertically-stacked fluid venting systems in the Middle Miocene to Holocene succession. Individual fluid-venting structures that form part of vertical venting systems include conical pockmarks, fluid related shallow depressions with flat bottom, chimney structures and positive high amplitude anomalies (PHAAs). Detailed seismic interpretation reveals for the first time that chimneys and PHAAs have a variety of plan form geometries (circular through to linear) within a given vertical succession of fluid venting structures. Linear chimneys are often associated with PHAAs which are interpreted as deposits of methane-related carbonate. The geometry and depth of depressions associated with fluid venting structures are used to infer relative rates of fluid flux or intensity of the fluid eruption. This classification scheme is as follows; linear PHAAs and conduits (slow fluid venting), sub-circular PHAAs and shallow depressions (slow to moderate rates of venting), pockmarks (fast rates of fluid venting). Two new types of pockmarks are identified based on the architecture of the sediments which infill them. They include advancing pockmark arrays and nested pockmarks. In contrast to normal pockmarks which are stacked vertically, successions of nested pockmarks and advancing pockmark arrays are laterally offset and migrate laterally, typically downslope. The reactivated craters of advancing pockmarks erode the downstream margin of preceding and underlying infill sequences whilst the infill sequence of nested pockmarks migrate gently downslope but without eroding the underlying and preceding infill sequence. Nested pockmarks and advancing pockmark arrays are confined to inclined surfaces. Downslope migration is a product of the interplay between slope inclination, sedimentation rate and bottom current activity. The trails of nested pockmarks and advancing pockmarks cluster above the axis of gas-bearing turbidite channels. PHAAs, chimneys, a present day bottom simulating reflector and negative high amplitude pockmark infills also occur in these areas and pockmarks occur above crestal faults which root in underlying rollover anticline. This implies the fluid source was derived from depth in the turtle anticline structure. The presence of negative, high-amplitude pockmark infills may suggest the fluid source was gas. A detailed spatial analysis and characterization of fluid venting structures on successive horizons in the middle Miocene to Holocene succession indicates that their distribution and type are affected by tectonic structures and vertical changes in the nature of the host sediments. Linear chimneys occur vertically below Linear PHAAs. The former are occur within the polygonally faulted interval whilst the later occur at the top of above the polygonally faulted interval. They tend to cluster in parts of the basin where the orientations of polygonal faults are strongly perturbed such as around salt diapirs and in salt-withdrawal synclines. Both linear venting structures are interpreted to post-date polygonal fault growth. Linear chimneys and linear PHAAs both have a close spatial and geometric relationship with PFs and deeper extending salt-related faults. The parallel relationship between linear venting structures and adjacent faults (salt or compaction related) are attributed to development and alignment of vertical hydraulic fractures (vertical conduits for linear chimneys) in the local fault induced stress field which subsequently provides fluid migration pathways. A model of vertical fluid migration through the polygonally faulted interval is proposed it involved initial fault-bound trapping, sealing and overpressure beneath an impermeable horizon in the lower part of the PF tier in the early stages, and vertical breaching, hydraulic fracturing and vertical fluid rise through the upper part of the tier in the later stages. Vertical changes in the morphology and type of fluid venting structures occur across small vertical transitions which reflect changes in gross lithology from fine-grained hemipelagites to chaotic and heterogeneous mass transport deposits (MTDs). A linear zone of positive high amplitude anomalies, referred to as a linear venting network, transitions to an array of elongate-to-sub circular shallow depressions with flat bases or conical pockmarks at the upper surface of MTDs. Further changes occur above MTDs where honeycomb pockmarks, so-called based on their hexagonal-shaped perimeter which coincides with polygonal fault intersections, exist. In these cases vertical changes in the type of fluid venting structures are attributed to contrasting patterns of mechanical failure in different sediment when subject to fluid overpressure. Although individual linear chimneys and PHAAs post-dating polygonal growth are strongly affected by the location and orientation of PFs those which precede polygonal fault growth such as pockmarks can affect the orientation of PFs. For example deep pockmark craters with the steepest sidewall inclinations coincide with overlying concentrically aligned PFs yet those which are shallower and have gently dipping sidewalls coincide with more isotropic PF patterns. This suggests that the topographic relief of pockmark craters or compaction above craters of certain depths perturb the stress state within sediments where polygonal faults form. This thesis has demonstrated that seismic interpretation of vertical successions of different types of fluid venting structures can be used to reconstruct spatial variations in the intensity of fluid flow at different stages in the evolution of basins.

Abstract Zircon in mudstones, sandstones and greenschist facies metasediments has been investigated using conventional SEM techniques (BSE, CL and SE imaging) and reveals highly variable microstructures and textures. In these rocks, zircon readily responds to low temperature events due to radiation damage in its crystal lattice while crystalline zircon remains unmodified. Zircon that alters as a result of metamictization has a low BSE intensity (dark BSE zircon) and electron microprobe measurements show an enrichment of Mg, Al, Ca, Fe, Y and a loss of Si and Zr, while Hf appears to remain relatively constant between the unmodified parent and the resulting modified phase. Dark BSE zircon forms via two main mechanisms. The dominant dark BSE zircon form (Group 1) has a microstructure containing an abundance of pores, cavities and inclusions forming as a result of a coupled dissolution-reprecipitation mechanism. Electron backscattered diffraction (EBSD) analyses suggests that the structure is composed of randomly orientated nanocrystalline zircon. The other form of dark BSE zircon (Group 2) is produced by a solid-state diffusion-driven cation-exchange process in which structural recovery occurs (as determined by EBSD) and where inclusions or pores are absent in the microstructure of the product phase. The different forms of altered zircon are chemically indistinct. Zircon outgrowths (c.1µm thick) on the margins of detrital unmodified zircon in clay-rich sandstones indicate Zr was transported from altered grains. These zircons crystallised below 100°C. The upper temperature at which metamict zircon may be dissolved is constrained by the annealing temperature of the zircon lattice whereby metamict areas are repaired above c.250°C. Zircon outgrowths are larger (c.3µm thick) and in much greater abundance in slates that have experienced deformation and temperatures c.350°C. They have a complex microstructure, partly as a result of interactions with xenotime that also forms outgrowths upto 12µm thick on zircon. Xenotime inclusions and zircon-xenotime complexes have been identified within zircon outgrowths using TEM and LV-STEM. Electron transparent foils of the outgrowth were prepared using the FIB microscope. Zircon outgrowths have similar chemical characteristics to dark BSE zircon but have distinctly different substitution relationships. There are also differences between the chemistry of dark BSE zircon from sedimentary and greenschist facies rocks where the latter is considerably more enriched in Al, Y and Ca. The implications of this are that zircon chemistry is strongly influenced by the local conditions in which in formed. Sedimentary processing causes considerable bias in the zircon population. Fine-grained sediment is a sink for high U and Th, heavily radiation damaged, old zircon in comparison to mature sediment that is likely to contain an abundance of low U and Th zircon or young zircon upon deposition. Small metamict zircon fragments are prone to dissolution and can be readily stripped from the matrix of fine-grained sediments. The concentration of zircon outgrowths in fine-grained sedimentary and metasedimentary rocks is primarily the result of sedimentological processes. The findings of this work illustrate the importance of studying minerals in-situ and within their petrological context. The wide-spread and abundant nature of low temperature zircon is a major consideration for geochronology, sedimentary provenance studies, the interpretation of zoning in zircon and has significant implications for the long-term storage of radioactive material.

The diagenetic history of the Oolite Group from Pwll du to Blaen Onneu has been unravelled. Cement types, solution events, dolomitisation, concretions and other authigenic minerals are described. The calcite cement crystals show chemical zonation through staining but cathodoluminescence (CL) has proved the most useful method for displaying internal features of calcite crystals. CL makes it possible to trace the growth of crystals and monitor changes in crystallographic form during growth. Seven cement zones are identified using stained specimens. Each zone has a characteristic luminescence. The age of these zones relative to the exposure of the top of the Oolite Group is established. Two distinct diagenetic areas are recognised on the basis of the distribution of the cement zones. Area 1 : the Pwll-y-cwm and Blaen Onneu Oolites in the Clydach area, in which Zones 1, 2b and 6 are present, Area 2 : the rest of the Oolite Group, in which Zones 2a, 3, 4, 5 and 6 are present. It is possible to correlate the cement zones of Area 2 along the outcrop for a distance of 8 km and also with cements in the overlying Llanelly formation. Oolomitisation prevents correlation of cements in Area 1 over a distance of more than 1/2 km. Using CL it is possible to identify solution surfaces, on a micron to millimetre scale, within the cements. four solution events are identified in Area 2. Solution affects that are previously unreported are described. The presence of solution surfaces is used to illustrate the constantly changing nature of the pore waters that have affected the Oolite Group. Carbon and oxygen stable isotope analysis of allochems and cement zones highlights the distinction between Areas 1 and 2 and are used to try and identify the nature of the pore waters affecting the Oolite Group during its diagenesis. The carbon isotopic composition of the allochems and cements pre-dating the exposure of the top of the Oolite Group is related to their distance below the exposure surface at the top of the Oolite Group and thus seems to reflect meteoric alteration. The successive cement zones do not show a progressive trend in caroon and oxygen values; however, in different samples successive cement zones do show the same pattern of changes in carbon and oxygen values. The values typical of a specific cement zone are thought to relate to the pore fluids from which it was precipitated. The pattern of changes seen in successive cement zones is explained in terms of changing pore water chemistry and isotopic composition with time. These changes occurred over an area at least the size of the study area. Very light oxygen values in veins are attributed to increased temperatures associated with deep burial. An attempt is made to reconstruct the paleohydrology of the study area in an attempt to explain the nature and distribution of the cements.

This thesis documents an investigation of selected intrusions of the Palaeocene granites on the Isle of Skye, and includes: petrography, mineral chemistry (include REE concentrations) fluid inclusions, whole rock major-, trace- and rare-earth-elements, together with stable oxygen and hydrogen and Sr-Nd isotope geochemistry on whole rocks and mineral separates. The data presented indicate that the granites are geochemically related and have been severely affected by 'associated' hydrothermal fluid(s), especially the Loch Ainort Granite. The hydrothermal fluids most likely represent the limited mixing between magmatic and meteoric waters in Beinn an Dubhaich Granite. In contrast, the other granites are likely to have been affected by meteoric water alone, with water-rock interaction having occurred below 500°C.The coarsening of granophyric texture outwards from alkali feldspar phenocrysts, the low concentration of Fe3+ in pyroxenes, together with the slight variations in Fe2+: Mg ratio of biotites, suggest a slow cooling history for the granites. The (two feldspar) Beinn an Dubhaich and Glamaig granites have small negative Eu anomalies compared with the (single feldspar) Loch Ainort and Marsco granites, suggesting the important role of plagioclase fractionation during granite evolution. The most important reservoirs for the REE in the Skye granites are amphibole, pyroxene and apatite. The inflection present in the plot of MnO against SiO2, the non-linear and divergent relationships between K2O, Rb, Sr and ?Nb against SiO2, and the presence of an anomalous group of granites which are characterised by low Zr, Y and Nb, but with high SiO2 contents, mitigates against simple mixing as the dominant process responsible for the compositional diversity of the granites.

In order to avoid the damaging climatic consequences of rising atmospheric CO2, and reduce current atmospheric CO2 concentrations to pre-industrial levels, anthropogenic CO2 emissions must be mitigated by capturing CO2 at power plants and storing it for thousands of years. Underground storage within deep geological formations, such as depleted gas and oil fields or deep saline aquifers, is the best understood solution for storage of CO2. In order for this method to gain more public and political acceptance it is important to characterise the potential causes, quantities and rates of CO2 release that could result if leakage were to occur from anthropogenic storage projects. This study examines two sites in the Colorado Plateau where faulted and actively leaking CO2 reservoirs provide natural analogues for failed anthropogenic storage sites. The two sites in question, the Little Grand Wash and northern Salt Wash graben faults are situated at the northern end of the Paradox Basin in Utah and represent classic three way traps due to juxtaposition of the shallow, north plunging Green River anticline against a set of east-west trending normal faults. In addition to active leakage sites in each area there are numerous fossilised travertine deposits. Along the Little Grand Wash fault the ancient mounds are restricted to the fault trace whereas ancient travertine mounds associated with the northern fault of the Salt Wash graben are far more numerous and occur up to ~530 m into the footwall of the fault. This more diffuse pattern of flow is due to the outcropping of unconfined aquifer units at the surface. A total of 45 U-Th dates from the majority of these travertine mounds provides a unique data set. The oldest deposits from the Little Grand Wash and northern Salt Wash graben faults produced ages of 113,912 ± 604 and 413,474 ± 15,127 years respectively. Repeat ages show reasonable reproducibility and analytical errors on results are of the order of 1% of the ages. The coupling of travertine elevation measurements with their radiometric ages gives an incision rate for each site. A rate of 0.342 m/ka for the Little Grand Wash fault relates directly to Green River incision and agrees with previous work on the Colorado Plateau, providing a further data point for characterisation of uplift of the province. For the northern fault of the Salt Wash graben a rate of 0.168 m/ka for the tributaries running through the area gives a robust method with which to estimate ages for un-dated mounds. The results of radiometric dating and incision rate age estimation of travertine mounds shows that leakage can last for timescales of 100,000’s of years, while high resolution U-Th dating of an individual mound demonstrated that leakage from a single point can last for a minimum of ~11,000 years. A range of travertine ages show that leakage to the surface has constantly switched location through time, while the presence of three mounds of distinct age at one location demonstrate that pathways can become repeatedly re-used over periods of ~45,000 years. There is no evidence of temporal periodicity in travertine deposition but there is a distinct spatial pattern of leakage as shown by localised similarities in the initial uranium chemistries of travertine mounds. Initial leakage is proximally located to the axial trace of the Green River anticline and subsequent leakage spreads from this central point along the fault plane in both east and west directions. The switching of fluid flow pathways to the surface can be explained by three main mechanisms: mineralisation, 3-phase interference of CO2 related fluid flow and seismically triggered alteration in dynamic strain acting upon the hydrology of the faults. These mechanisms have differing influences in each area - demonstrating that the behaviour of fluid flow switching in a system confined to damage zone fractures (Little Grand Wash fault) is different to a system leaking through an unconfined aquifer (northern fault of the Salt Wash graben). Coupling of travertine ages with estimates of their volumes provided a total worse case scenario for quantity of CO2 leakage of 6.2 x 10^6 ± 1.7 x 10^6 tonnes for the Little Grand Wash fault and 7.4 x 10^6 ± 2 x 10^6 tonnes for the northern fault of the Salt Wash graben. From these totals time averaged leakage rates of 55 ± 15 and 47 ± 13 tonnes/year were estimated for each fault. The leakage rate for the actively precipitating Crystal Geyser travertine (which is the result of anthropogenic exploration drilling) is estimated to be 3,153 ± 851 tonnes/year. These total and modern rates provide analogues for leakage via caprock failure and catastrophic wellbore failure. Applying them to large scale storage sites such as Weyburn and Gorgon revealed that for caprock failure complete leakage of these reservoirs will take place over timescales of 10^5-10^6 years, while for catastrophic failure of a single well complete leakage of these reservoirs could occur over as little as 10^3 – 10^4 years. This finding has important implications for the successful monitoring of anthropogenic storage sites.

The compound eyes of trilobites, which appeared in the Early Cambrian, represent one of the first preserved visual systems. Application of state-of-the-art microscopy techniques in the present study has revealed fine details of the microstructure and chemistry of these unusual calcite eyes that, until now, have been inaccessible and this has facilitated new insights into their growth and function. Six species from three families of trilobite with holochroal eyes, ranging from Early Ordovician to Middle Carboniferous, and 21 species from three families of trilobite with schizochroal eyes, ranging from Early Ordovician to Middle Devonian, were investigated. High-resolution microscopy techniques, including Electron Backscatter Diffraction, Transmission Electron Microscopy and Electron Probe Micro-analysis have made it possible to ‘see’ through the diagenesis of trilobite lenses to reveal the likely original lens microstructure and chemistry. Computer-based optical modelling has further shown how original lens microstructure and chemistry enhanced lens function. The discovery of sub-micron sized crystals that display a gradual and precise change in orientation shows that in many lenses much of the original microstructure has remained intact despite exposure to pore fluids and elevated temperatures and pressures during diagenesis. Although microstructure varies slightly with lens shape, there is often exceptionally precise crystal orientation. In holochroal eyes there is a direct relationship between lens shape and microstructure; where lens surfaces are planar, crystals are of uniform orientation, but where lens surfaces are convex, c axis orientation fans out, away from the lens axis. Several microstructural patterns have been identified in schizochroal lenses. However, a single original microstructural pattern, in which c axis orientation fans out at lens surfaces but remains parallel to the lens axis in the centre of the lens, may be applicable to all schizochroal lenses. The original chemical composition of the lenses, in particular those of schizochroal type, is less commonly preserved. However, the unravelling of a diagenetic pathway of change, through understanding the intricacies of relationships between different minerals in the lenses, has made possible a better understanding of how these lenses were altered during diagenesis. Lenses in holochroal eyes are invariably low-magnesium calcite, like the rest of the exoskeleton, as has been established for some time. The present study clarifies the original structures of schizochroal lenses and in doing so, ends the controversy over lens function; lenses were originally constructed as doublets, as suggested by Clarkson and Levi-Setti, and were not gradient index lenses, as was suggested by Campbell and Bruton and Haas. Lenses in schizochroal eyes, were constructed of high-magnesium calcite, with highest concentrations of magnesium in the lower ‘intralensar bowl’ and central ‘core’ regions of the lens. The degree of partitioning of high levels of magnesium, of up to 8 mole % MgCO3, in the schizochroal eyes is remarkable given the magnesium-poor ‘calcite seas’ in which they were formed. This is perhaps the first example of element partitioning within biominerals for a specific function. Based on the growth sequence of modern arthropod exoskeletons lenses in trilobite eyes are likely to have grown from the outer surface in, one lamella at a time, with microstructure and chemical composition controlled by an organic matrix. Assessment of the trilobite optical structures, using Code V optical modelling software, leads to the conclusions that the trilobite eyes functioned in a similar manner to the apposition eyes of modern animals. Code V modelling of holochroal and schizochroal eyes, and the subsequent determination of their resolution and sensitivity, shows that both eye types probably had a single optically isolated photoreceptor beneath each lens. Using a combination of specific lens size, shape, spacing, microstructure and composition, the schizochroal eye was adapted to low light intensities, similar to the eye of the modern isopod Cirolana. These adaptations would have provided the trilobite with light and dark detection of a resolution sufficient to identify movement, allowing it to detect prey and defend itself against predators. The birefringent properties of the calcite from which these lenses were made could be a hindrance, resulting in double refraction of light rays and the formation of ‘ghost’ images. Fascinatingly however this property provides the lenses with the refractive power required to make full use of the light available to them, vital for an organism with a crystalline lens with a fixed focal length. Study of the calcified lenses of ostracods and brittlestars and comparison to lenses in schizochroal trilobite eyes confirms that these modern organisms do not provide accurate analogues for trilobite eyes. No other organism that shares all characteristics of schizochroal trilobite eyes has yet been found; the eyes of the phacopine trilobites remain unique in the natural world.

Earthquake ruptures along tectonically active faults nucleate predominantly at depths of 5 to 12km in the crust, so the portions of faults that slip in these events cannot be directly observed. The geometry and composition of seismogenic faults controls the nucleation, propagation and termination of the earthquake rupture process. This study aims to place constraints on the geometry and composition of seismogenic faults by examining ancient faults exhumed from the depths at which earthquakes are observed to nucleate. Faults exposed in the Sierra Nevada, California, show that the internal architecture of earthquake faults is heterogeneous at a variety of scales. Field and microstructural observations are used to describe in detail the architecture of two pseudotachylyte-bearing fault systems in the Granite Pass region of Sequoia and Kings Canyon National Park; the Granite Pass fault (GPF) and associated faults, and the Glacier Lakes fault (GLF) and faults that splay from the GLF. The GPF and sub-parallel faults are 1 to 6.7km long with left-lateral strike-slip displacements up to 80m. The GPF and GPF-parallel faults have architectures that are heterogeneous along strike. They are composed of one to four fault core strands containing cataclasites and ultracataclasites that cross-cut early localized crystal-plastic deformation. Slip surfaces developed at the edges of, within and between fault cores are defined by pseudotachylytes and cataclasites with thicknesses of ~0.01 to 20mm. Fault-related subsidiary structures are developed on either side of fault cores, and comprise damage zones with widths orthogonal to the fault of up to 30m. The GLF and splay faults have architectures that are more homogeneous along strike. These faults are composed of a tabular volume of heavily fractured and altered host rock between approximately planar fault core strands. The fault cores are centimetres wide and contain cataclasites and foliated cataclasites that are cross-cut by pseudotachylytes. Fault-related damage is limited in extent to several metres beyond the bounding fault cores. The GLF contains additional cataclasites, ultracataclasites and pseudotachylytes in a fault core strand within the tabular zone of fractured rock. Thermochronologic analyses of the host rock granodiorite, combined with previously published palaeogeobarometry and apatite fission track data, define the temperature and pressure changes associated with cooling and exhumation of the pluton. The P-T conditions prevalent during the deformation history of the GPF fault system are evaluated by relating recrystallisation mechanisms in quartz to temperature, showing that the early stages of deformation occurred at temperatures of 450 to 600ºC. Dating of pseudotachylytes by the K-Ar isotopic method suggests subsequent brittle deformation took place at temperatures <350ºC and pressures ≤150MPa. A model for the architecture of the GPF architecture therefore has well constrained environmental controls, and should be transferrable to faults with comparable deformation histories. Small faults (cumulative displacements <1m) in the Mount Abbot Quadrangle, 55km north of Granite Pass, have been examined to illustrate the processes associated with the earliest stages of slip in the Sierra Nevada faults. The faults have branched or straight fault traces. Pseudotachylytes in branching faults show that these faults accumulated displacement in high velocity slip events, rather than by quasi-static fault growth. Branching faults without pseudotachylytes contain chlorite breccias interpreted to have formed in response to slip along faults with elevated pore fluid pressure. Straight faults also likely underwent slip events, but contain cataclased chlorite and epidote, suggesting low fluid pressures during slip. The small faults show that fluid-rock interactions are critical to fault geometry, and that lateral structural heterogeneity is established after small finite displacements. Field and thin section observations of exhumed seismogenic faults show that fault architecture and fault rock assemblage are critical to the earthquake rupture process. The heterogeneous composition of slip surfaces in the GPF faults imply that melt lubrication cannot account for all of the dynamic slip weakening as there are no continuous pseudotachylyte generation surfaces through the fault zones. Multiple slip weakening mechanisms must have been active during single rupture events. Slip weakening mechanisms also change at a given point on the fault in response to continued deformation. Splay faults at the GLF termination suggest that structural complexity observed at the terminations of fault surface traces can also be expected at depth. The off-fault damage at the termination of the GLF will change the bulk elastic properties of the host rock and must be accounted for in models of rupture propagation beyond fault terminations, or across geometrical discontinuities. Additionally, aftershock distributions and focal mechanisms may be controlled by the geometry of structures present at fault terminations.