We apply a spherical harmonic analysis to the Point Source Redshift Survey (PSCz), to compute the real-space galaxy power spectrum and the degree of redshift distortion caused by peculiar velocities. We employ new parameter eigenvector and hierarchical data compression techniques, allowing a much larger number of harmonic modes to be included, and correspondingly smaller error bars. Using 4644 harmonic modes, compressed to 2278, we find that the IRAS redshift-space distortion parameter is $\beta = 0.39 \pm 0.12$ and the amplitude of galaxy clustering on a scale of $k=0.1 \Mpch$ is $\Delta_{\rm gal}(0.1)=0.42 \pm 0.02$. Combining these we find the amplitude of mass perturbations is $\Delta_m(0.1)=(0.16\pm0.04) \Omega_m^{-0.6}$. A preliminary model fitting analysis combining the PSCz amplitudes with the CMB and abundance of clusters yields the cosmological matter density parameter $\Omega_m=0.16\pm 0.03$, the amplitude of primordial perturbations $Q=(8.4\pm 3.8) \times 10^{-5}$, and the IRAS bias parameter $b=0.84\pm 0.28$.

We have constrained the redshift-distortion parameter $\beta \equiv \Omega^{0.6}/b$ and the real-space power spectrum of the IRAS PSCz survey using a spherical-harmonic redshift-distortion analysis combined with a data compression method which is designed to deal with correlated parameters. Our latest result, $\beta=0.4 \pm 0.1$, strongly rules out $\beta=1$.

The star formation history of the universe shows strong evolution with cosmological epoch. Although we know mergers between galaxies can cause luminous bursts of star formation, the relative importance of such mergers to the global star formation rate (SFR) is unknown. We present a simple analytic formula for the rate at which halos merge to form higher-mass systems, derived from Press-Schechter theory and confirmed by numerical simulations (for high halo masses). A comparison of the evolution in halo formation rate with the observed evolution in the global SFR indicates that the latter is largely driven by halo mergers at z>1. Recent numerical simulations by Kolatt et al. (1999) and Knebe & Muller (1999) show how merging systems are strongly biased tracers of mass fluctuations, thereby explaining the strong clustering observed for Lyman-break galaxies without any need to assume that Lyman-break galaxies are associated only with the most massive systems at z~3.

We apply the formalism of spherical harmonic decomposition to the galaxy density field of the IRAS PSCz redshift survey. The PSCz redshift survey has almost all-sky coverage and includes IRAS galaxies to a flux limit of 0.6 Jy. Using maximum likelihood methods to examine (to first order) the distortion of the galaxy pattern due to redshift coordinates, we have measured the parameter \beta= \Omega^{0.6}/b. We also simultaneously measure (a) the undistorted amplitude of perturbations in the galaxy distribution when a parameterised power spectrum is assumed, or (b) the shape and amplitude of the real-space power spectrum if the band-power in a set of passbands is measured in a step- wise fashion. These methods are extensively tested on a series of CDM, \Lambda CDM and MDM simulations and are found to be unbiased. We obtain consistent results for the subset of the PSCz catalogue with flux above 0.75 Jy but inclusion of galaxies to the formal flux limit of the catalogue gives variations which are larger than our internal errors. For the 0.75 Jy catalogue we find, in the case of a parameterised power spectrum, \beta=0.58\pm0.26 and the amplitude of the real space power measured at wavenumber k=0.1 h/Mpc is \Delta_{0.1}=0.42 \pm 0.03. Freeing the shape of the power spectrum we find that \beta=0.47\pm 0.16 (conditional error), and \Delta_{0.1}=0.47 \pm 0.03. The shape of the real-space power spectrum is consistent with a \Gamma= 0.2 CDM-like shape parameter, but does not strongly rule out a number of other models. Finally by combining our estimate of the amplitude of galaxy clustering and the distortion parameter we find the amplitude of mass fluctuations on a scale k=0.1 \hmpc is \Delta_\rho = 0.24 \Omega_0^{-0.6}$, with an uncertainty of 50%.

It has been proposed that the cosmological constant $\Lambda$ might be measured from geometric effects on large-scale structure. A positive vacuum density leads to correlation-function contours which are squashed in the radial direction when calculated assuming a matter-dominated model. We show that this effect will be somewhat harder to detect than previous calculations have suggested: the squashing factor is likely to be $<1.3$, given realistic constraints on the matter contribution to $\Omega$. Moreover, the geometrical distortion risks being confused with the redshift-space distortions caused by the peculiar velocities associated with the growth of galaxy clustering. These depend on the density and bias parameters via the combination $\beta\equiv \Omega^{0.6}/b$, and we show that the main practical effect of a geometrical flattening factor $F$ is to simulate gravitational instability with $\beta_{\rm eff}\simeq 0.5(F-1)$. Nevertheless, with datasets of sufficient size it is possible to distinguish the two effects; we discuss in detail how this should be done. New-generation redshift surveys of galaxies and quasars are potentially capable of detecting a non-zero vacuum density, if it exists at a cosmologically interesting level.