| This thesis presents the results of applying rigorous methods of statistical inference to the evaluation and interpretation of the spectral line candidates exhibited by the gamma-ray burst (GRB) data of GB870303, observed by the Japanese satellite Ginga. This work has fundamental importance in the current debate over the distance scale of GRBs: it provides strong evidence that at least some GRBs occur in our galaxy.; I begin by evaluating the evidence for lines in the GB870303 data. These data exhibit dips in two spectra, denoted S1 and S2, separated by 22.5 s. S1, spanning 4 s, exhibits a single dip at {dollar}approx{dollar}20 keV, while S2, spanning 9 s, exhibits harmonically-spaced dips at {dollar}approx{dollar}20 and 40 keV. I employ frequentist and Bayesian statistical methodologies to evaluate these dips. I find that the significances of the continuum-plus-line(s) models requested by the S1, S2, and combined (S1+S2) data are 2.2 {dollar}times{dollar} 10{dollar}sp{lcub}-5{rcub}{dollar}, 4.0 {dollar}times{dollar} 10{dollar}sp{lcub}-4{rcub}{dollar} and 1.1 {dollar}times{dollar} 10{dollar}sp{lcub}-6{rcub}{dollar} respectively, and I find that the odds favoring the continuum-plus-line(s) models are 66:1, 2.1:1, and 235:1, respectively.; I then demonstrate that these data are consistent with the hypothesis that lines are formed by cyclotron resonant scattering in the circumstellar environment of a Galactic neutron star. I model line formation with a Monte Carlo code, injecting continuum photons into a static plane-parallel slab of electrons threaded by a superstrong magnetic field (B {dollar}sim{dollar} 10{dollar}sp{lcub}12{rcub}{dollar} G) oriented with angle {dollar}Psi{dollar} relative to the slab normal. I examine two slab geometries, which differ in where photons are injected into the slab. I compare azimuthally-symmetric models ({dollar}Psi{dollar} = 0), which I posit represent line formation at the magnetic pole, with models having more general field orientations. I find that the S1 and S2 data are consistent with both slab geometries, and with all values of {dollar}Psi{dollar}, with the exception that for the semi-infinite atmosphere slab geometry, the preferred location for the formation of the S1 line is the magnetic equator. The combined (S1+S2) data show marginal evidence for line formation at the equator, and for neutron star rotation, on which I place limits. |
