Abstract
This thesis provides new measurements the Mg isotopic abundances using high resolution
spectroscopy of quasar absorption systems. Magnesium plays a central role
in many recent measurements of varying fine structure constant, α. The 2796/2803
doublet is important because it is commonly detected and is an anchor line , relatively
insensitive to alpha variation, providing a reference point for more sensitive
transitions.
In the absence of reliable high redshift measurements, previous measurements
assumed terrestrial isotopic abundances. This assumption is critical: if incorrect and
if the relative isotopic abundances evolve with redshift, as expected, it could mimic a
redshift dependent alpha.
Two non-linear least-squares methods were used. The first is a parabolic interpolation
method and the second solves directly for the Mg isotope strengths. We study
154 absorbers, ranging in redshift from zem = 0.2 to 2.4. Using parabolic interpolation
we derive 93 measurements, finding a mean abundance of the primary isotope
24Mg= 57±7% (compared to terrestrial of 79%). Using least-squares optimisation, we
obtain measurements for 133 systems. When fitted assuming no alpha variation, we
find a mean abundance 24Mg = 60±2%. Allowing for alpha variation, we derive 143
measurements, yielding a mean abundance 24Mg= 55±2%.
We investigate the spatial variation suggested by King et al. (2012) and for a
dipole-only model find a dipole direction RA = (17.5±0.6)hr, dec. = (−49±8)deg, significant
at 4.4σ. When fitting a dipole+monopole model, we find the previously statistically
significant monopole term vanishes.
Our results agree with 24Mg = 60% suggested by King et al. (2012) to explain the
monopole term. We thus suggest future varying alpha analyses, where Mg is used,
should allow for the results presented here.
We also study the impact of quasar slit-centroiding variations on 114 measurements.
Discarding systems potentially susceptible to the effect reduces the sample
size from 279 to 114 absorption systems. Re-fitting a dipole-only model to this subset
produces a dipole direction of RA = (18.7±1.1)hr, dec. = (−50±14), significant at 2.3σ.