We present high resolution optical and UV spectra of the 46 min orbital period, helium binary, GP Com. Our data contains simultaneous photometric correction which confirms the flaring behaviour observed in previous optical and UV data. In this system all lines show a triple peaked structure where the outer two peaks are associated with the accretion disc around the compact object. The main aim of this paper is to constrain the origin of the central peak, also called ``central spike''. We find that the central spike contributes to the flare spectra indicating that its origin is probably the compact object. We also detect that the central spike moves with orbital phase following an S-wave pattern. The radial velocity semiamplitude of the S-wave is ~10 km/s indicating that its origin is near the centre of mass of the system, which in this case lies very close to the white dwarf. Our resolution is higher than that of previous data which allows us to resolve structure in the central peak of the line. The central spike in three of the HeI lines shows another peak blueshifted with respect to the main peak. We propose that one of the peaks is a neutral helium forbidden transition excited in a high electron density region. This forbidden transition is associated with the permitted one (the stronger peak in two of the lines). The presence of a high electron density region again favours the white dwarf as their origin.

Subdwarf B (sdB) stars are thought to be helium burning stars with low mass hydrogen envelopes. Several evolutionary paths have been proposed to explain the formation of these systems. One of these scenarios is the evolution of the sdB progenitor within a binary system. We have looked systematically at bright sdB stars from the PG survey. By taking spectra at several different epochs we have measured the radial velocity shifts caused by the motion of the sdB star within the binary. Our data have been taken over a long time base line (2 years) which allowed us to find longer period binaries than known before. Here we present results for 29 sdB systems.

Subdwarf B (sdB) stars are thought to be core helium burning stars with low mass hydrogen envelopes. In recent years it has become clear that many sdB stars lose their hydrogen through interaction with a binary companion and continue to reside in binary systems today. In this paper we present the results of a programme to measure orbital parameters of binary sdB stars. We determine the orbits of 22 binary sdB stars from 424 radial velocity measurements, raising the sample of sdBs with known orbital parameters to 38. We calculate lower limits for the masses of the companions of the sdB stars which, when combined with the orbital periods of the systems, allow us to discuss approximate evolutionary constraints. We find that a formation path for sdB stars consisting of mass transfer at the tip of the red giant branch followed by a common envelope phase explains most, but not all of the observed systems. It is particularly difficult to explain both long period systems and short period, massive systems. We present new measurements of the effective temperature, surface density and surface helium abundance for some of the sdB stars by fitting their blue spectra. We find that two of them (PG0839+399 and KPD1946+4340) do not lie in the Extreme Horizontal Branch (EHB) band indicating that they are post-EHB stars.

We present lightcurves of four binary subdwarf B stars, Ton 245, Feige 11, PG1432+159 and PG1017-086. We also present new spectroscopic data for PG1017-086 from which we derive its orbital period, P=0.073d, and the mass function, f_m = 0.0010 +/- 0.0002 solar masses. This is the shortest period for an sdB binary measured to-date. The values of P and f_m for the other sdB binaries have been published elsewhere. We are able to exclude the possibility that the unseen companion stars to Ton 245, Feige 11 and PG1432+159 are main-sequence stars or sub-giant stars from the absence of a sinusoidal signal which would be caused by the irradiation of such a companion star, i.e., they show no reflection effect. The unseen companion stars in these binaries are likely to be white dwarf stars. By contrast, the reflection effect in PG1017-086 is clearly seen. The lack of eclipses in this binary combined with other data suggests that the companion is a low mass M-dwarf or, perhaps, a brown dwarf.

Although a large number of orbital periods of cataclysmic variable stars (CVs) have been measured, comparison of period and luminosity distributions with evolutionary theory is affected by strong selection effects. A test has been discovered which is independent of these selection effects and is based upon the kinematics of CVs (Kolb & Stehle, 1996). If the standard models of evolution are correct then long-period (P_orb > 5 hrs) CVs should be typically less than 1.5 Gyr old, and their line-of-sight velocity dispersion ($\sigma_\gamma$) should be small. We present results from a pilot study which indicate that this postulate is indeed true. Four long-period dwarf novae (EM Cyg, V426 Oph, SS Cyg and AH Her) were observed over a complete orbit, in order that accurate radial velocities be obtained. We find values of -1.7, 5.4, 15.4 and 1.8 km/s with uncertainties of order 3 km/s, referred to the dynamical Local Standard of Rest (LSR), leading to a dispersion of ~ 8 km/s. Calculation of a 95 per cent confidence interval gives the result 4 < \sigma_\gamma < 28 km/s compared to a prediction of 15 km/s. We also have an improved determination of mass donor spectral type, K_2 and q for the four systems.

(Abbreviated) We have used precise radial velocity measurements of subdwarf-B stars from the Palomar-Green catalogue to look for binary extreme horizontal branch (EHB) stars. We identify 36 EHB stars in our sample and find that at least 21 of these stars are binaries. All but one or two of these are new identifications. The minimum binary fraction for EHB stars implied by our survey is 60+-8%. Our survey is sensitive to binaries with orbital periods P less than about 10d. For reasonable assumptions concerning the period distribution and the mass ratio distribution of the binaries, we find that the mean detection efficiency of our survey over this range of orbital periods is 87%. Allowing for this estimated detection efficiency, the fraction of EHB stars which are short-period binaries ($0.03 < P <10d, approximately) is 69+-9%. The value is not strongly dependent on the period distribution below P=10d or the mean companion mass for these short-period binaries. The orbital separation of the stars in these binaries is much less than the size of the red giant from which the EHB star has formed. This is strong evidence that binary star evolution is fundamental to the formation of the majority of EHB stars. If there are also binary EHB stars whose orbital periods are more than about 10d, the fraction of EHB stars whose evolution has been affected by the presence of a companion may be much higher.

We present high resolution spectra of the AM CVn helium binary GP Com at two different wavelength ranges. The spectra show the same flaring behaviour observed in previous UV and optical data. We find that the central spike contributes to the flare spectra indicating that its origin is probably the compact object. We also detect that the central spike moves with orbital phase following an S-wave pattern. The radial velocity semiamplitude of the S-wave is \~10 km/s which indicates its origin is near the centre of mass of the system, which in this case lies very close to the white dwarf. The Stark effect seems to affect significantly the central spike of some of the lines suggesting that it forms in a high electron density region. This again favours the idea that the central spike originates in the white dwarf. We present Doppler maps obtained for the emission lines which show three clear emission regions.

We present spectra of the pulsating sdB star KPD1930+2752 which confirm that this star is a binary. The radial velocities measured from the H-alpha and HeI6678 spectral lines vary sinusoidally with the same period (2h 17m) as the ellipsoidal variability seen by Billeres et al. (2000). The amplitude of the orbital motion (349.3+-2.7 km/s) combined with the canonical mass for sdB stars (0.5 solar masses) implies a total mass for the binary of 1.47+-0.01 solar masses The unseen companion star is almost certainly a white dwarf star. The binary will merge within about 200 million years due to gravitational wave radiation. The accretion of helium and other elements heavier than hydrogen onto the white dwarf which then exceeds the Chandrasekhar mass (1.4 solar masses) is a viable model for the cause of Type Ia supernovae. KPD1930+2752 is the first star to be discovered which is a good candidate for the progenitor of a Type Ia supernova of this type which will merge on an astrophysically interesting timescale.

We have discovered that the spectrum of the well-known dwarf nova EM Cyg is contaminated by light from a K2-5V star (in addition to the K-type mass donor star). The K2-5V star contributes approximately 16 per cent of the light from the system and if not taken into account has a considerable effect upon radial velocity measurements of the mass donor star. We obtain a new radial velocity amplitude for the mass donor star of K2 = 202 +/- 3 km/s, which compares with the value of K2 = 135 +/- 3 km/s obtained in Stover, Robinson & Nather's classic 1981 study of EM Cyg. The revised value of the amplitude combined with a measurement of rotational broadening of the mass donor vsini = 140 +/- 6 km/s, leads to a new mass ratio of q = M2/M1 = 0.88 +/- 0.05. This solves a long standing problem with EM Cyg because Stover et al.'s measurements indicated a mass ratio q > 1, a value which should have led to dynamically unstable mass transfer for the secondary mass deduced by Stover et al. The revised value of the mass ratio combined with the orbital inclination i = 67 +/- 2 degrees leads to masses of 0.99 +/- 0.12 Msun and 1.12 +/- 0.08 Msun for the mass donor and white dwarf respectively. The mass donor is evolved, since it has a later spectral type (K3) than its mass would imply. We discuss whether the K star could be physically associated with EM Cyg or not, and present the results of the spectroscopic study.