Long Baseline Interferometry is a well established technique that enables high angular resolution measurements to be made with a radio interferometer containing independent local oscillators and signal recorders. This Thesis addresses the problems encountered when this technique is applied to low frequency (81.5 MHz) astronomy, and specifically methods of improving its phase stability. The technique is used to reassess existing evidence for the existence of large-scale structure associated with the quasar 3C48, and to give a much better understanding of the structure and behaviour of the supernova remnant Cassiopeia A at low frequencies. Much of the instrumentation used for this work already existed, but in a form unsuitable for measuring phase. The first section of the Thesis shows how these data collection and digital correlator systems could be modified and extended to improve sensitivity and generate true interferometric phase. Two phase calibration schemes are also considered. Firstly, the ‘thin triangle’ method, which uses the closure phase generated by a three-element interferometer to determine the true astronomical phase of a source comprising a single, bright point source and an associated resolvable component. This includes an analysis of how best to determine closure phase when the signal-to-noise ratio on one or more of the baselines is low, and shows techniques based on the ‘triple-product’ method, proposed by Cornwell, to perform best. Secondly, a new method of instrumental stabilisation is introduced and demonstrated, in which the interferometer is calibrated by broadband signals from VHF public broadcast transmitters, which are processed along with the astronomical data. The technique is shown to be capable of removing practically all the instrumental phase drifts usually associated with a long baseline interferometer on baselines up to 100 km. A number of new observations are also presented, mostly made with the stabilised mobile interferometer described above. Firstly of the bright quasar 3C48 to investigate earlier reports of an associated steep-spectrum extended component. These show the importance of considering the effects of confusion in the earlier observations and prompt a critical reassessment of the existing evidence, which is shown to be flawed. Detailed observations of the supernova remnant Cassiopeia A are also presented, with a resolution and sensitivity previously unattained at metre wavelengths. Comparisons with 20 cm VLA data of the same epoch show the remnant to look remarkably similar at the two frequencies, though tentative spectral index measurements indicate that its younger features have slightly steeper spectra than the rest. Furthermore, sensitive observations made on a 128 km baseline show that Cassiopeia A does not contain the steep-spectrum compact component with a flux density of more than 5 Jy reported by many earlier authors. An even deeper search at 408 MHz, this time for pulsed emission from the remnant, puts an upper limit of 80 mJy on the flux density due to any associated pulsar with a period 4 > P > 0.02 sec.