Search

Legal witnesses are perceived to be a crucial part of international criminal tribunals' and courts' responses to mass human rights violations, not only in contributing to a legal determination of guilty or not guilty being reached but also to produce a collective memory of the atrocities. However, this common claim in the transitional justice legal scholarship fails to fully understand the nature of memory. Memory construction entails fragments of individual and collective memories combining into contingent and contestable narratives of the past, and it is for this reason that the thesis challenges the claim that international criminal tribunals and courts are able to produce a collective memory of atrocities. Taking the International Criminal Tribunal for Rwanda (ICTR) as its case study, this thesis differs from the majority of legal transitional justice studies researching courts and tribunals, which have worked within traditional legal frameworks. The originality of this thesis is that it offers a conceptually driven and empirically grounded analysis of archived ICTR documents, and interview transcripts of ICTR staff from the University of Washington (UoW) archive, relating to the selection of witnesses. In order to show the limitations of legal memory the thesis constructs an original conceptual framework using Giorgio Agamben's concept of 'witness' and Paul Ricoeur's philosophical insights on 'memory'. Accordingly, the research argues for the need to understand witnessing as non-instrumental: a contingent and multi-layered discursive process. It is the discursive formations within a discourse of witnessing that constitutes the conditions determining the subject position of witness. Therefore, what the witness subject remembers is a constellation of fragments of the past, which are formed within the contingent conditions of discourse. The arguments advanced by the thesis contribute towards discussions on the role of witnesses and how the past is remembered during transitional periods, by highlighting the need for the study and practice of transitional justice to fully understand how theoretical insights can make visible the complexities and contours of the legal construction of the past. The thesis thus contributes to specific analysis of witnessing and memory, but also the broader transitional justice scholarship.

It is still not yet understood how structure in the Universe was seeded. The theory of inflation provides a framework to understand the origins of structure as quantum fluctuations of the vacuum, whilst maintaining an attractive simplicity, and giving a natural explanation of how the Universe is so flat, homogeneous, and isotropic. If inflation occurred, it would have produced a gravitational wave background, indirectly observable today through its effect on the polarization of the cosmic microwave background. We present a new model for the polarized microwave sky, combining data from the Planck and WMAP satellites, and legacy radio surveys, with theoretical models for Galactic foreground SEDs. We present an example use-case of this model by forecasting the effect of Galactic foregrounds on a future satellite mission. We then move on to consider an extension to the simplest inflationary scenario in which an additional set of SU(2) gauge fields are present during inflation, and may source a chiral gravitational wave background. We show that future CMB missions will not be able to distinguish such a scenario from the simplest single field slow-roll model using standard two- point statistical techniques, and that space-based laser interferometers will provide a complementary probe by constraining the short wavelength behavior of the blue-tilted gravitational wave spectrum. Finally, we demonstrate the efficacy of a new method to account for complex Galactic foregrounds in the analysis of high-sensitivity polarized CMB data from the upcoming Simons Observatory. We first show that ignoring the spatial variation could lead to a false detection of a gravitational wave background, and that this can be corrected for by introducing additional parameters describing the spatial variation of the foreground spectral energy dependence.

The necessity for increasing the efficiency of combustion engines whilst simultaneously reducing the emissions of various pollutants is widely accepted. With reference specifically to premixed spark ignition (SI) engines, the turbulence/flame interaction is of particular relevance as it not only determines the maximum possible burning rate in such an engine, but also if flames are likely to extinguish (quench). The present study describes a three-dimensional laser imaging technique which has been developed to study turbulent premixed flames in a fan-stirred combustion bomb. This has, for the first time, allowed the full 3D structure of flames developing at engine-like conditions to be analysed without the assumptions required in previous works using 2D flame imaging techniques. Methane/air mixtures at a starting temperature and pressure of 298K and 0.1 MPa respectively and an equivalence ratio with respect to stoichiometry of 0.6 were centrally ignited using a laser ignition system at various root-mean-square (RMS) turbulence velocities, where for each explosion, a 3D image of the growing flame was captured at various instants in time. The flame images were quantitatively characterised to yield flame surface area, burned gas volume, reaction progress variable and flame surface density. From the flame surface areas and burned gas volumes, turbulent burning velocities were directly measured from surface area ratios and compared with correlations of turbulent burning velocities made previously at Leeds using 2D flame imaging techniques. This tested the assumption that flame structure observed with 2D imaging techniques gave a good representation of the overall 3D flame structure. An excellent agreement was found between the datasets, indicating that such assumptions were valid under the conditions employed in the present work. The surface area ratio data obtained in the present work were also compared with the predictions of direct numerical simulation (DNS) studies conducted at Cambridge University, where it was found that the experimentally derived results typically exhibited slightly higher flame surface area ratios. This was attributed to the combustion chamber geometry in the DNS study and differences in mixture properties between the two studies, with more work at similar starting conditions required to elucidate the exact cause of the differences. Reaction progress variable was calculated for the 2D laser sheet images obtained in the present work and their 3D counterparts. Although a good agreement was seen where the central laser sheet image in a 3D reconstruction was analysed, disparities were observed when laser sheet images at a fixed position in the combustion bomb was used at high root mean square turbulence velocities. This indicates that the use of fixed position laser sheet imaging techniques, such as particle image velocimetry, are of limited use for highly turbulent flame analysis.