We analyze the mean rest-frame ultraviolet (UV) spectrum of Type Ia Supernovae(SNe) and its dispersion using high signal-to-noise Keck-I/LRIS-B spectroscopyfor a sample of 36 events at intermediate redshift (z=0.5) discoveredby the Canada-France-Hawaii Telescope Supernova Legacy Survey (SNLS). Weintroduce a new method for removing host galaxy contamination in our spectra,exploiting the comprehensive photometric coverage of the SNLS SNe and theirhost galaxies, thereby providing the first quantitative view of the UV spectralproperties of a large sample of distant SNe Ia. Although the mean SN Ia spectrumhas not evolved significantly over the past 40 percent of cosmic history, preciseevolutionary constraints are limited by the absence of a comparable sample ofhigh quality local spectra. The mean UV spectrum of our z 0.5 SNe Ia and itsdispersion is tabulated for use in future applications. Within the high-redshiftsample, we discover significant UV spectral variations and exclude dust extinctionas the primary cause by examining trends with the optical SN color. Although progenitor metallicity may drive some of these trends, the variations we see aremuch larger than predicted in recent models and do not follow expected patterns.An interesting new result is a variation seen in the wavelength of selected UVfeatures with phase. We also demonstrate systematic differences in the SN Iaspectral features with SN lightcurve width in both the UV and the optical. Weshow that these intrinsic variations could represent a statistical limitation in thefuture use of high-redshift SNe Ia for precision cosmology. We conclude thatfurther detailed studies are needed, both locally and at moderate redshift wherethe rest-frame UV can be studied precisely, in order that future missions canconfidently be planned to fully exploit SNe Ia as cosmological probes.

R-band intensity measurements along the light curve of Type Ia supernovae discovered by the Cosmology Project (SCP) are fitted in brightness to templates allowing a free parameter the time-axis width factor w identically equal to s times (1+z). The data points are then individually aligned in the time-axis, normalized and K-corrected back to the rest frame, after which the nearly 1300 normalized intensity measurements are found to lie on a well-determined common rest-frame B-band curve which we call the "composite curve." The same procedure is applied to 18 low-redshift Calan/Tololo SNe with Z < 0.11; these nearly 300 B-band photometry points are found to lie on the composite curve equally well. The SCP search technique produces several measurements before maximum light for each supernova. We demonstrate that the linear stretch factor, s, which parameterizes the light-curve timescale appears independent of z, and applies equally well to the declining and rising parts of the light curve. In fact, the B band template that best fits this composite curve fits the individual supernova photometry data when stretched by a factor s with chi 2/DoF ~;~; 1, thus as well as any parameterization can, given the current data sets. The measurement of the data of explosion, however, is model dependent and not tightly constrained by the current data. We also demonstrate the 1 + z light-cure time-axis broadening expected from cosmological expansion. This argues strongly against alternative explanations, such as tired light, for the redshift of distant objects.

The Nearby Supernova Factory (SNfactory) is an ambitious project to find and study in detail approximately 300 nearby Type Ia supernovae (SNe Ia) at redshifts 0.03 < z < 0.08. This program will provide an exceptional data set of well-studied SNe in the nearby smooth Hubble flow that can be used as calibration for the current and future programs designed to use SNe to measure the cosmological parameters. The first key ingredient for this program is a reliable supply of Hubble-flow SNe systematically discovered in unprecedented numbers using the same techniques as those used in distant SNe searches. In 2002, 35 SNe were found using our test-bed pipeline for automated SN search and discovery. The pipeline uses images from the asteroid search conducted by the Near Earth Asteroid Tracking group at JPL. Improvements in our subtraction techniques and analysis have allowed us to increase our effective SN discovery rate to ~;12 SNe/month in 2003.