Abandoning the assumption that light traces mass, I seek the location of the center of the Local Group of galaxies based solely on kinematic data and the plausible assumption of infall. The available set of positions and radial velocities is shown to be a misleading indicator of Local Group motions, giving a direction to the center offset from the true one; statistical techniques of moderate sophistication do not catch the offset. Corrected calculations show the center to lie in the direction to M31 within the uncertainty of the method, a few degrees. The distance to the center is not well determined, lying about 0.5 Mpc from the Milky Way. The pattern of observed (galactocentric) radial velocities excludes both dynamically important `orphan haloes' and any extended dark matter halo for the Group as a whole, and shows the Group to have formed from a much more extended volume than it presently occupies. Kinematics alone indicates that the mass of the Group is concentrated effectively in M31 and the Milky Way.

A significant amount of data on the historical and current behaviour of variable stars is derived from visual estimates of brightness using a set of comparison stars. To make optimum use of this invaluable collection one must understand the characteristics of visual photometry, which are significantly different from those of electronic or photographic data. Here I show that the dispersion of estimates among observers is very consistent at between 0.2 and 0.3 magnitudes and, surprisingly, has no apparent dependence on the colour of comparison stars or on their spacing in brightness.

In a pioneering exposition of mathematical astronomy for the public, Sir John Herschel attributed the stability of the ring of Saturn to its being eccentric with respect to the planet and lopsided (asymmetric in mass) by a minute amount. Tracing the sources and effects of this error reveals several lessons of general relevance to science: on the formulation and interpretation of calculations, the use of cutting-edge observations and the combining of observations with theory. I emphasise the phenomenon of reinforcing errors.

A simplified formula for gravitational-radiation power is examined. It is shown to give completely erroneous answers in three situations, making it useless even for rough estimates. It is emphasized that short timescales, as well as fast speeds, make classical approximations to relativistic calculations untenable.

We present the results of an all-sky, deep optical survey for faint Local Group dwarf galaxies. Candidate objects were selected from the second Palomar survey (POSS-II) and ESO/SRC survey plates and follow-up observations performed to determine whether they were indeed overlooked members of the Local Group. Only two galaxies (Antlia and Cetus) were discovered this way out of 206 candidates. Based on internal and external comparisons, we estimate that our visual survey is more than 77% complete for objects larger than one arc minute in size and with a surface brightness greater than an extremely faint limit over the 72% of the sky not obstructed by the Milky Way. Our limit of sensitivity cannot be calculated exactly, but is certainly fainter than 25 magnitudes per square arc second in R, probably 25.5 and possibly approaching 26. We conclude that there are at most one or two Local Group dwarf galaxies fitting our observational criteria still undiscovered in the clear part of the sky, and a roughly a dozen hidden behind the Milky Way. Our work places the "missing satellite problem" on a firm quantitative observational basis. We present detailed data on all our candidates, including surface brightness measurements.

In 1925 Sir James Jeans calculated that a star made up of an ideal gas, generating energy as a moderately positive function of temperature and density, could not exist. Such stars would be unstable to oscillations of increasing size. It appears that the flaw in his calculation has never been clearly explained, especially the physical basis for it. I conclude it lies in an almost offhand assumption made about the form of the temperature perturbation. The episode has lessons concerning the structure of stars, the use of perturbation methods (especially in discussing stability), the use of mathematical models and the process by which one theory is chosen over its competitors.

A beginning end-point for galaxy motions within the 10-Mpc Local Volume is constructed by requiring a smooth distribution of (luminous) mass at the time of recombination, which is shown to be equivalent to a smooth Hubble flow at early times. It is found, by this purely kinematical method, that present peculiar motions are too small by a factor of at least several (and largely in the wrong direction) to have produced the observed structures within the age of the universe. Known dynamical effects are inadequate to remove the discrepancy. This result is different in origin from previously known ``cold flow'' problems. The simple dynamical picture often used within the Local Volume (for instance, in deriving masses through calculation of a zero-velocity surface) is thus called into question. The most straightforward explanation (though not the only possible) is that there exists a large quantity of baryonic matter in this region so far undetected, and unassociated with galaxies or groups.

We present low resolution optical spectroscopy for six cataclysmic variable candidates and the old nova V888 Cen. We confirm the classification as cataclysmic variable for LB 9963 and FQ Mon, while the other four candidates turn out to be different type of stars. We discuss the individual spectra and pay special attention to the mass transfer rate and disc temperature and density of the three cataclysmic variables.

Using high-quality data on 149 galaxies within 10 megaparsecs (Mpc), I find no correlation between luminosity and peculiar velocity at all. There is no unequivocal sign on scales of 1-2 Mpc of the expected gravitational effect of the brightest galaxies, in particular infall toward groups; or of infall toward the Supergalactic Plane on any scale. Either dark matter is not distributed in the same way as luminous matter in this region, or peculiar velocities are not due to fluctuations in mass. The sensitivity of peculiar velocity studies to the background model is highlighted.

The meaning of the expansion of the universe, or the `expansion of space,' is explored using two phenomena: the motion of a test particle against a homogeneous background and the cosmological redshift. Contrary to some expectations, a particle removed from the Hubble flow never returns to it. The cosmological redshift is not an `expansion effect;' in special cases it can be separated into a kinematic (special relativistic) part and a static (gravitational redshift) part, but in general must be thought of as the effect on light rays of their propagation through curved space-time. Space as such does not affect matter by `expanding,' but only through its curvature.

The kinematics of galaxies with 10 megaparsecs (10 Mpc) of the Milky Way is investigated using published distances and radial velocities. With respect to the average Hubble flow (isotropic or simple anisotropic), there is NO systematic relation between peculiar velocity dispersion and absolute magnitude over a range of 10 magnitudes; neither is there any apparent variation with galaxy type or between field and cluster members. There are several possible explanations for the lack of variation, though all have difficulties: either there is no relationship between light and mass on these scales, or the peculiar velocities are not produced by gravitational interaction, or the background dynamical picture is wrong in some systematic way. The extremely cold local flow of 40-60 km/s dispersion reported by some authors is shown to be an artifact of sparse data, a velocity dispersion of over 100 km/s being closer to the actual value. Galaxies with a high (positive) radial velocity have been selected against in studies of this volume, biasing numerical results.

We present observations of 82 Local Group dwarf galaxy candidates, of which 62 were chosen visually from ESO-SRC survey plates of the southern sky (32 of which were not previously catalogued) and the rest suggested by various sources in the literature. Two are the Local Group galaxies Antlia and Cetus; nine are more distant galaxies, though still within a few megaparsecs; 45 are background galaxies; seven are planetary (or other emission) nebulae; 15 are reflection or other Galactic nebulae; two are galaxy clusters; one is a Galactic star cluster and one is a misidentified star. We conclude that there is no large population of faint Local Group dwarf galaxies of any familiar type awaiting discovery. We point out the danger of relying on a single type of data to reach conclusions about an object.

Formulating the equations of motion for cosmological bodies (such as galaxies) in an integral, rather than differential, form has several advantages. Using an integral the mathematical instability at early times is avoided and the boundary conditions of the integral correspond closely with available data. Here it is shown that such a least-action calculation for a number of bodies interacting by gravity has a finite number of solutions, possibly only one. Characteristics of the different possible solutions are explored. The results are extended to cover the motion of a continuous fluid. A method to generalize an action to use velocities, instead of positions, in boundary conditions, is given, which reduces in particular cases to those given by Giavalisco et al. (1993) and Schmoldt & Saha (1998). The velocity boundary condition is shown to have no effect on the number of solutions.

We report the discovery of new member of the Local Group in the constellation of Antlia. Optically the system appears to be a typical dwarf spheroidal galaxy of type dE3.5 with no apparent young blue stars or unusual features. A color-magnitude diagram in I, V-I shows the tip of the red giant branch, giving a distance modulus of 25.3 +/- 0.2 (1.15 Mpc +/- 0.1) and a metallicity of -1.6 +/- 0.3. Although Antlia is in a relatively isolated part of the Local Group it is only 1.2 degrees away on the sky from the Local Group dwarf NGC3109, and may be an associated system.