| Millisecond pulsars make excellent targets for pulse timing observations. Times-of-arrival (TOA's) can be measured to microsecond accuracies, and are stable over several years of observations. This permits unparalleled precision in astrometry, time-keeping, solar system dynamics, gravitation physics, and close binary evolution. Millisecond pulsars may also provide the most sensitive system for detection of gravitational waves with periods of a few years.; This thesis describes the development of our third-generation timing system, and its use at Arecibo Observatory. This system utilizes four real-time coherent dedispersers, which remove pulse smearing caused by interstellar dispersion. The data taken using the dedispersers along with that from regular filter bank spectrometers provide the most precise pulsar TOA's obtainable.; To yield any physics, raw TOA's must be precisely referenced to terrestrial time standards, and transformed to an inertial reference frame located at the solar system barycenter using an accurate model of the earth's rotation and motion. The barycentric TOA's must then be fit into a model of the pulsar's spin, position, proper motion, and binary motion. The submicrosecond precision of our measurements puts stringent requirements on each of these steps, and provides some of the most accurate astrometric and binary measurements available, including the measurement of annual parallax and general relativistic effects that allow us to measure the mass of pulsar. The differences between the predicted and measured TOA's can themselves be used, for example, to constrain the energy density of long wavelength gravitational radiation.; The arrival times from our most precise timing measurements, those of PSR 1937 + 21, clearly show deviations from the model far greater than our measurement errors, with greater power at very long wavelengths. The analysis of this steep-spectrum noise cannot be done with standard Fourier algorithms. We have developed a spectral estimation technique that is a valid estimator of the true power spectrum and useful with unevenly sampled data. We use the measured spectrum to characterize the likely sources of the timing noise and to place a limit on the cosmic gravitational wave background near frequencies of 0.2 cycles yr{dollar}sp{lcub}-1{rcub}{dollar} of {dollar}Omegasb g{dollar} {dollar}<{dollar} 1 {dollar}times{dollar} 10{dollar}sp{lcub}-7{rcub}{dollar} (95% confidence). |
