The HiRes detector represents a step forward in the use of the air fluorescence technique in the study of extremely high energy cosmic rays (discussed in chapter 1). It builds upon the success of the Fly's Eye detector and will detect larger numbers of high energy extensive air showers at greater distances and finer resolution. To achieve these aims has required the development and testing of new detector units, and has brought new problems and challenges.

This thesis presents work performed by the author on the prototype HiRes detector described in chapter 3. Whilst much of the development of HiRes was based on both past experience with the Fly's Eye detector and the use of Monte Carlo simulations, the author has sought to demonstrate that we have developed the necessary expertise to solve many of the expected problems on {\em experimental} data. However on such a large project, it becomes necessary to limit one's scope to a subset of all the problems and tasks associated with the development of a new detector. With this in mind the author has concentrated on the development and application of high precision timing information to allow the coherent use of information from both detector sites.

In chapter 4 the author details the development of Global Positioning System (GPS) based clocks to provide high precision relative timing accuracy between the two HiRes sites. As a result of this work, the relative timing accuracy of the two sites is less than 50ns the majority of the time, with occasional degradations in the timing accuracy to around 100ns. Absolute timing accuracy is within 340ns of UTC 95% of the time. As part of this work the timing system was checked and calibrated giving us confidence in our ability to correctly assign triggering times of PMTs.

The implementation of high precision relative timing accuracy between the two HiRes sites allowed the development of a stereo-timing air shower reconstruction program which is discussed in chapter 5. The lack of high precision relative timing accuracy had previously prevented the use of this technique on the Fly's Eye detectors. However Monte Carlo simulations had indicated that if one had high precision relative timing between the two sites, and one could account for all timing systematics, then the technique should provide reconstruction accuracy better than any other method in use. Thus the author undertook the development and testing of a stereo-timing reconstruction program.

Testing and optimisation of the reconstruction program utilised the Laserscope, a portable telescope mounted UV laser which can be steered and fired to produce a beam of UV light in the atmosphere capable of triggering the HiRes detectors. The Laserscope thus provided a large dataset of cosmic ray like tracks with known positions and trajectories. The Laserscope data was also used in attempts to parameterise PMT trigger time slewing, a known source of systematic error in assigning PMT triggering times. The results of this work indicated that for EAS observed with opening angles greater than about 10 degrees (where the opening angle is the angle between the two shower detector planes) the median reconstruction error was 0.4 degrees with 95% of events being reconstructed with errors of less than 1 degree. For EAS viewed at small opening angles, the reconstruction accuracy is worse, with a median error of 0.8 degrees and 95% limit of 1.8 degrees. An evaluation of the performance of the stereo-timing reconstruction program on data from the stage 1.0 detector configuration was also performed by reconstructing showers using only the lowest elevation ring of mirrors at HiRes 1 (and the two elevation rings of mirrors at HiRes 2). Encouragingly results indicated only a small increase in the reconstruction error despite the loss of 54 degrees of elevation angle coverage between the two configurations.

High precision relative timing accuracy between the two HiRes sites allowed the development of an accurate stereo-timing reconstruction program. An accurately reconstructed trajectory is the important first step in determining EAS parameters such as depth of shower maximum and primary particle energy. These parameters are estimated through the use of a program to compare the measured light profile of the EAS with that expected assuming a Gaisser-Hillas shower development function. Chapter 6 thus describes the compilation of the stereo dataset and the development of a profile fitting program for use on this dataset.

Fitting the measured light profile to a Gaisser-Hillas function is a well understood problem. However application to the HiRes stereo dataset brings with it new problems. EAS are now being viewed at much larger distances than those detected by the Fly's Eye. The effect of the atmosphere, and in particular aerosol scattering and attenuation of light, is a much more significant problem than it was for the Fly's Eye. This has long been recognised, so that nightly monitoring of atmospheric conditions is performed.

These nightly measurements were incorporated into the profile fitting program, and a comparison made of the use of different aerosol models. The most significant atmospheric effect is attenuation due to aerosol scattering of light, which is partly characterised by an aerosol extinction length. The comparison indicated that the use of the Fly's Eye fixed aerosol extinction length tends to significantly overestimate the primary particle energy. This was due to measured extinction lengths being consistently longer than the Fly's Eye value, which, due to the significantly larger distances involved with the detection of light by HiRes 2, results in over estimation of primary particle energy. Care must be taken with the use of these nightly measurements to ensure that they are not biased by the presence of ground fog or an aerosol mixing layer. Evaluation of different aerosol scattering phase functions was more difficult, with the comparison being largely insensitive to the different phase functions considered. However it was noted that energy estimates using a phase function based on a log normal distribution of particles were systematically higher than those using the Fly's Eyes phase function which is based on a power law distribution of aerosol particles. Previous work indicates that a phase function based on a log normal distribution of particles should be more appropriate, so until further evidence is obtained preference should probably be given to a lognormal based phase function.

From analysis of the stereo dataset, it appears that the HiRes detector is achieving the desired aims. We are able to well reconstruct EAS at very large distances from the detector. Energy and depth of shower maximum distributions appear reasonable, with the expected increase in aperture with energy being observed. The performance of event reconstruction assuming a stage 1.0 configuration was compared with results using the prototype configuration. This comparison indicated that, provided the region around shower maximum was observed, the energy estimates should be reasonable. Given the elevation angle coverage of the stage 1.0 detector this condition will only be met for the more distant (and in general more energetic) showers.