Search

We measure the number of companions per galaxy (Nc) as a function of r-band absolute magnitude for both the Sloan Digital Sky Survey and the Croton et al. (2006) semi-analytic catalog applied to the Millennium Run simulation. For close pairs with projected separations of 5-20 h^{-1} kpc, velocity differences less than 500 km/s, and luminosity ratios between 1:2 and 2:1, we find good agreement between the observations and simulations, with Nc consistently close to 0.02 over the range -22 < M_r < -18. For larger pair separations, Nc(M_r) instead becomes increasingly steep towards the faint end, implying that luminosity-dependent clustering plays an important role on small scales. Using the simulations to assess and correct for projection effects, we infer that the real-space Nc(M_r) for close pairs peaks at about M*, and declines by at least a factor of two as M_r becomes fainter. Conversely, by measuring the number density of close companions, we estimate that at least 90% of all major mergers occur between galaxies which are fainter than L*. Finally, measurements of the luminosity density of close companions indicate that L* galaxies likely dominate in terms of the overall importance of major mergers in the evolution of galaxy populations at low redshift., Comment: Accepted for publication in ApJ. 20 pages, including 9 figures

We compare the use of galaxy asymmetry and pair proximity for measuring galaxy merger fractions and rates for a volume limited sample of 3184 galaxies with -21 < M(B) -5 log h < -18 mag. and 0.010 < z < 0.123 drawn from the Millennium Galaxy Catalogue. Our findings are that: (i) Galaxies in close pairs are generally more asymmetric than isolated galaxies and the degree of asymmetry increases for closer pairs. At least 35% of close pairs (with projected separation of less than 20 h^{-1} kpc and velocity difference of less than 500 km s^{-1}) show significant asymmetry and are therefore likely to be physically bound. (ii) Among asymmetric galaxies, we find that at least 80% are either interacting systems or merger remnants. However, a significant fraction of galaxies initially identified as asymmetric are contaminated by nearby stars or are fragmented by the source extraction algorithm. Merger rates calculated via asymmetry indices need careful attention in order to remove the above sources of contamination, but are very reliable once this is carried out. (iii) Close pairs and asymmetries represent two complementary methods of measuring the merger rate. Galaxies in close pairs identify future mergers, occurring within the dynamical friction timescale, while asymmetries are sensitive to the immediate pre-merger phase and identify remnants. (iv) The merger fraction derived via the close pair fraction and asymmetries is about 2% for a merger rate of (5.2 +- 1.0) 10^{-4} h^3 Mpc^{-3} Gyr^{-1}. These results are marginally consistent with theoretical simulations (depending on the merger time-scale), but imply a flat evolution of the merger rate with redshift up to z ~1., Comment: 10 pages, 10 figures, emulateapj format. ApJ, accepted

We compare the structural properties of two classes of galaxies at intermediate redshift: those in dynamically close galaxy pairs, and those which are isolated. Both samples are selected from the CNOC2 Redshift Survey, and have redshifts in the range 0.1 < z <0.6. Hubble Space Telescope WFPC2 images were acquired as part of a snapshot survey, and were used to measure bulge fraction and asymmetry for these galaxies. We find that paired and isolated galaxies have identical distributions of bulge fractions. Conversely, we find that paired galaxies are much more likely to be asymmetric (R_T+R_A >= 0.13) than isolated galaxies. Assuming that half of these pairs are unlikely to be close enough to merge, we estimate that 40% +/- 11% of merging galaxies are asymmetric, compared with 9% +/- 3% of isolated galaxies. The difference is even more striking for strongly asymmetric (R_T+R_A >= 0.16) galaxies: 25% +/- 8% for merging galaxies versus 1% +/- 1% for isolated galaxies. We find that strongly asymmetric paired galaxies are very blue, with rest-frame B-R colors close to 0.80, compared with a mean (B-R)_0 of 1.24 for all paired galaxies. In addition, asymmetric galaxies in pairs have strong [OII]3727 emission lines. We conclude that close to half of the galaxy pairs in our sample are in the process of merging, and that most of these mergers are accompanied by triggered star formation., Comment: Accepted for publication in the Astronomical Journal. 40 pages, including 15 figures. For full resolution version, please see http://www.trentu.ca/physics/dpatton/hstpairs/

We investigate redshift evolution in the galaxy merger and accretion rates, using a well-defined sample of 4184 galaxies with 0.12 < z < 0.55 and R_C < 21.5. We identify 88 galaxies in close (5 < r_p < 20 h^{-1} kpc) dynamical (delta v < 500 km/s) pairs. These galaxies are used to compute global pair statistics, after accounting for selection effects resulting from the flux limit, k-corrections, luminosity evolution, and spectroscopic incompleteness. We find that the number of companions per galaxy (for -21 < M_B^{k,e} < -18) is Nc = 0.0321 +/- 0.0077 at z=0.3. The luminosity in companions, per galaxy, is Lc = 0.0294 +/- 0.0084 x 10^10 h^2 L_sun. We assume that Nc is proportional to the galaxy merger rate, while Lc is directly related to the mass accretion rate. After increasing the maximum pair separation to 50 h^{-1} kpc, and comparing with the low redshift SSRS2 pairs sample, we infer evolution in the galaxy merger and accretion rates of (1+z)^{2.3 +/- 0.7} and (1+z)^{2.3 +/- 0.9} respectively. These are the first such estimates to be made using only confirmed dynamical pairs. When combined with several additional assumptions, this implies that approximately 15% of present epoch galaxies with -21 < M_B < -18 have undergone a major merger since z=1., Comment: 40 pages, 8 figures. Accepted for publication in the Astrophysical Journal

(abridged) The systematic variation of galaxy colors and types with clustering environment could either be the result of local conditions at formation or subsequent environmental effects as larger scale structures draw together galaxies whose stellar mass is largely in place. At z~0.4 the co-moving galaxy correlation length, r_0, measured in the CNOC2 sample is strongly color dependent, rising from 2/h Mpc to nearly 10/h Mpc as the volume-limited subsamples range from blue to red. The luminosity dependence of r_0 at z~0.4 is weak below L_ast although there is an upturn at high luminosity where its interpretation depends on separating it from the r_0-color relation. The dominant effect of the group environment on star formation is seen in the radial gradient of the mean galaxy colors which on the average become redder than the field toward the group centers. The redder-than-field trend applies to groups with a line-of-sight velocity dispersion, sigma_1>150 kms. There is an indication, somewhat statistically insecure, that the high luminosity galaxies in groups with sigma_1<125 kms become bluer toward the group center. We conclude that the higher velocity dispersion groups largely act to suppress star formation relative to the less clustered field, leading to ``embalmed'' galaxies. The tidal fields within the groups appear to be a strong candidate for the physical source of the reduction of star formation in group galaxies relative to field. Tides operate effectively at all velocity dispersions to remove gas rich companions and low density gas in galactic halos. Given that much of the field population is in groups we suggest that this suppression may be the dominant galaxy evolution force at low redshift., Comment: ApJ accepted

We examine how the spatial correlation function of galaxies from the CNOC2 Field Galaxy Redshift Survey depends on galaxy color, luminosity and redshift. The projected correlation function w_p is determined for volume-limited samples of objects with 0.12 < z < 0.51 and evolution-compensated Rc absolute magnitudes M < -20, over the comoving projected separation range 0.04 Mpc/h < r_p < 10 Mpc/h. Our sample consists of 2937 galaxies which are classified as being either early- or late-type objects according to their spectral energy distribution (SED), determined from UBVRcIc photometry. For simplicity, galaxy SEDs are classified independently of redshift: our classification scheme therefore does not take into account the colour evolution of galaxies., Comment: 37 pages, 14 figures. To appear in The Astrophysical Journal

We report on the first measurement of the average mass and mass-to-light ratio of galaxy groups by analysing the weak lensing signal induced by these systems. The groups, which have velocity dispersions of 50-400 km/s, have been selected from the Canadian Network for Observational Cosmology Field Galaxy Redshift Survey (CNOC2). This survey allows the identification of a large number of groups with redshifts ranging from z=0.12-0.55, ideal for a weak lensing analysis of their mass distribution. For our analysis we use a sample of 50 groups which are selected on the basis of a careful dynamical analysis of group candidates. We detect a signal at the 99% confidence limit. The best fit singular isothermal sphere model yields an Einstein radius of 0.72+-0.29". This corresponds to a velocity dispersion of 274^{+48}_{-59} km/s (using photometric redshift distributions for the source galaxies), which is in good agreement with the dynamical estimate. Under the assumption that the light traces the mass, we find an average mass-to-light ratio of 191+-83 h Msun/Lsun in the restframe B band. Unlike dynamical estimates, this result is insensitive to problems associated with determining group membership. After correction of the observed mass-to-light ratio for luminosity evolution to z=0, we find 254+-110 h Msun/Lsun, lower than what is found for rich clusters. We use the observed mass-to-light ratio to estimate the matter density of the universe, for which we find Omega_m=0.19+-0.10 (Omega_Lambda=0), in good agreement with other recent estimates. For a closed universe, we obtain Omega_m=0.13+-0.07., Comment: Accepted for ApJ Letters 4 pages, 2 figures, uses emulateapj5.sty

Galaxy groups likely to be virialized are identified within the CNOC2 intermediate redshift galaxy survey using an iterative method. The number-velocity dispersion relation is in agreement with the low-mass extrapolation of the cluster normalized Press-Schechter function. The two-point group-group correlation function has r_0=6.8+/- 0.3 Mpc, which is larger than the correlations of individual galaxies at the level predicted from n-body calibrated halo clustering. The groups are stacked in velocity and position to create a sample large enough for measurement of a density and velocity dispersion profile. The stacked mean galaxy density profile falls nearly as a power law with r^{-2.5} and has no well-defined core. The projected velocity dispersion is examined for a variety of samples with different methods and found to be either flat or slowly rising outwards. The combination of a steeper-than-isothermal density profile and the outward rising velocity dispersion implies that the mass-to-light ratio of groups rises with radius. The M/L can be kept nearly constant if the galaxy orbits are nearly circular, although such strong tangential anisotropy is not supported by other evidence. The segregation of mass and light is not dependent on galaxy luminosity but is far more prominent in the red galaxies than the blue. The M/L gradient could arise from orbital ``sloshing'' of the galaxies in the group halos, dynamical friction acting on the galaxies in a background of ``classical'' collisionless dark matter, or, more speculatively, the dark matter may have a true core., Comment: 50 pages, 24 figures, submitted to ApJ

The Canadian Network for Observational Cosmology (CNOC2) Field Galaxy Redshift Survey is a spectroscopic/photometric survey of faint galaxies over 1.5 square degrees of sky with a nominal spectroscopic limit of R_c=21.5 mag. The primary goals of the survey are to investigate the evolution of galaxy clustering and galaxy populations over the redshift range of approximately 0.1 to 0.6. The survey area contains four widely separated patches on the sky with a total spectroscopic sample of over 6000 redshifts and a photometric sample of over 40,000 galaxies with 5-color photometry. We describe the survey and observational strategies, multi-object spectroscopy mask design procedure, and data reduction techniques for creating the spectroscopic-photometric catalogs. We also discuss the derivations of various statistical weights for the redshift sample which allow it to be used as a complete sample. As the initial release of the survey data, we present the data set and some statistics for the Patch CNOC0223+00., Comment: 28 pages, AAS-Latex version 5.0, 15 eps figures. Accepted for publication by the Astrophysical Journal Supplement

The rate of mass accumulation due to galaxy merging depends on the mass, density, and velocity distribution of galaxies in the near neighborhood of a host galaxy. The fractional luminosity in kinematic pairs combines all of these effects in a single estimator which is relatively insensitive to population evolution. Here we use a k-corrected and evolution compensated volume-limited sample drawing about 300 redshifts from CFGRS and 3000 from CNOC2 to measure the rate and redshift evolution of merging. We identify kinematic pairs with projected separations less than either 50 or 100 \hkpc and rest-frame velocity differences of less than 1000\kms. The fractional luminosity in pairs is modeled as f_L(Delta v,r_p,M_r^{ke})(1+z)^{m_L} where [f_L,m_L] are [0.14+/-0.07,0+/-1.4] and [0.37+/-0.7,0.1+/-0.5] for r_p<= 50 and 100\hkpc, respectively (Omega_M=0.2, Omega_Lambda=0.8). The value of m_L is about 0.6 larger if Lambda=0. To convert these redshift space statistics to a merger rate we use the data to derive a conversion factor to physical space pair density, a merger probability and a mean in-spiral time. The resulting mass accretion rate per galaxy (M_1,M_2>= 0.2 M*) is 0.02+/-0.01(1+z)^{0.1+/-0.5} M*~Gyr^{-1}. Present day high-luminosity galaxies therefore have accreted approximately 0.15M* of their mass over the approximately 7 Gyr to redshift one. (abridged), Comment: accepted for publication in ApJ Letters

We introduce two new pair statistics, which relate close galaxy pairs to the merger and accretion rates. We demonstrate the importance of correcting these (and other) pair statistics for selection effects related to sample depth and completeness. In particular, we highlight the severe bias that can result from the use of a flux-limited survey. The first statistic, denoted N_c, gives the number of companions per galaxy, within a specified range in absolute magnitude. N_c is directly related to the galaxy merger rate. The second statistic, called L_c, gives the total luminosity in companions, per galaxy. This quantity can be used to investigate the mass accretion rate. Both N_c and L_c are related to the galaxy correlation function and luminosity function in a straightforward manner. We outline techniques which account for various selection effects, and demonstrate the success of this approach using Monte Carlo simulations. If one assumes that clustering is independent of luminosity (which is appropriate for reasonable ranges in luminosity), then these statistics may be applied to flux-limited surveys. These techniques are applied to a sample of 5426 galaxies in the SSRS2 redshift survey. Using close dynamical pairs, we find N_c(-21

The redshift evolution of the galaxy two-point correlation function is a fundamental cosmological statistic. To identify similar galaxy populations at different redshifts, we select a strict volume-limited sample culled from the 6100 cataloged CNOC2 galaxies. Our high-luminosity subsample selects galaxies having k-corrected and evolution-compensated R luminosities, M_R^{k,e}, above -20 mag (H_0=100 km/s/Mpc) where M_ast^{k,e}(R)simeq -20.3 mag. This subsample contains about 2300 galaxies distributed between redshifts 0.1 and 0.65 spread over a total of 1.55 square degrees of sky. A similarly defined low-redshift sample is drawn from the Las Campanas Redshift Survey. We find that the co-moving two-point correlation function can be described as xi(r|z) = (r_00/r)^gamma (1+z)^{-(3+epsilon-gamma)} with r_{00}=5.03+/-0.08/h Mpc, epsilon=-0.17+/- 0.18 and gamma=1.87+/-0.07 over the z=0.03 to 0.65 redshift range, for Omega_M=0.2, Lambda=0. The measured clustering amplitude and its evolution are dependent on the adopted cosmology. The measured evolution rates for Omega_M=1 and flat Omega_M=0.2 background cosmologies are epsilon=0.80+/-0.22 and epsilon=-0.81+/-0.19, respectively, with r_{00} of 5.30+/-0.1/h Mpc and 4.85+/-0.1/h Mpc, respectively. The sensitivity of the derived correlations to the evolution corrections and details of the measurements is presented. The analytic prediction of biased clustering evolution for only the low density, LambdaCDM cosmology is readily consistent with the observations, with biased clustering in an open cosmology somewhat marginally excluded and a biased Omega_M=1 model predicting clustering evolution that is more than 6 standard deviations from the measured value., Comment: ApJ accepted. Analysis now done on updated catalogues leading to changes in details

The correlation evolution of a high luminosity subsample of the CNOC2 redshift survey is examined. The sample is restricted to galaxies for which the k corrected and evolution corrected R luminosity is M_R <=-20 mag, where M_* ~= -20.3 mag. This subsample contains about 2300 galaxies. In consort with 13000 galaxies in a similarly defined low redshift sample from the Las Campanas Redshift survey we find that the comoving correlation can be described as xi(r|z) = (r_00/r)^gamma (1+z)^{-(3+e)} with r_00=5.08 +/- 0.08/h Mpc, e=0.02 +/- 0.23 and gamma=1.81 +/- 0.03 over the z=0.03 to 0.65 redshift range in a cosmology with Omega_M=0.2, Lambda=0. The measured clustering amplitude, and its evolution, are dependent on the adopted cosmology. The evolution rates for Omega_M=1 and flat low density models are e=0.9 +/- 0.3 and e=-0.5 +/- 0.2, respectively, with r_00 ~= 5/h Mpc in all cases., Comment: To appear in the PASP proceedings of the conference: "The Hy-Redshift Universe: Galaxy Formation and Evolution at High Redshift", held in Berkeley, June 21-24, 1999. Conference Proceedings Editors: Andrew J. Bunker & Wil J. M. van Breughel

We examine the evolution of the galaxy luminosity function (LF) using a sample of over 2000 galaxies, with 0.12 < z < 0.55 and 17.0 < Rc < 21.5, drawn from the Canadian Network for Observational Cosmology Field Galaxy Redshift Survey (CNOC2), at present the largest such sample at intermediate redshifts. We use UBVRcIc photometry to classify our galaxies into early, intermediate, and late types, for which we compute luminosity functions in the rest-frame B, Rc, and U bandpasses. In particular, we adopt a parameterization of LF evolution including luminosity and number density evolution, and take care to quantify correlations among our LF evolution parameters. Our principal result is a clear quantitative separation of luminosity and density evolution for different galaxy populations, and the finding that the character of the LF evolution is strongly dependent on galaxy type. Specifically, the early- and intermediate-type LF's show primarily brightening at higher redshifts and only modest density evolution, while the late-type LF is best fit by strong number density increases at higher z, with little luminosity evolution. We also carefully measure and account for sample selection effects as functions of galaxy magnitude and color, and show that our results are not significantly affected by potential systematic effects, such as surface brightness selection, photometric errors, or redshift incompleteness. (abstract abridged), Comment: 57 pages, including 21 figures; ApJ, in press. Also available at http://manaslu.astro.utoronto.ca/~carlberg/cnoc/index.html

We report on the discovery of the radio, infrared and optical transient coincident with an X-ray transient proposed to be the afterglow of GRB 980703. At later times when the transient has faded below detection, we see an underlying galaxy with R=22.6; this galaxy is the brightest host galaxy (by nearly 2 magnitudes) of any cosmological GRB thus far. In keeping with an established trend, the GRB is not significantly offset from the host galaxy. Interpreting the multi-wavelength data in the framework of the popular fireball model requires that the synchrotron cooling break was between the optical and X-ray bands on July 8.5 UT and that the intrinsic extinction of the transient is Av=0.9. This is somewhat higher than the extinction for the galaxy as a whole, as estimated from spectroscopy., Comment: 5 pages, 3 figures, and 2 tables. Submitted to the Astrophysical Journal Letters on 27 August 1998

The CNOC2 field galaxy redshift survey is designed to provide measurements of the evolution of galaxies and their clustering over the redshift range 0 to 0.7. The sample is spread over four sky patches with a total area of about 1.5 square degrees. Here we report preliminary results based on two of the sky patches, and the redshift range of 0.15 to 0.55. We find that galaxy evolution can be statistically described as nearly pure luminosity evolution of early and intermediate SED types, and nearly pure density evolution of the late SED types. The correlation of blue galaxies relative to red galaxies is similar on large scales but drops by a factor of three on scales less than about 0.3/h mpc, approximately the mean scale of virialization. There is a clear, but small, 60%, change in clustering with 1.4 mag of luminosity. To minimize these population effects in our measurement of clustering evolution, we choose galaxies with M_r^{k,e}<= -20 mag as a population whose members are most likely to be conserved with redshift. Remarkably, the evolution of the clustered density in proper co-ordinates at r < 10/h Mpc, proportional to r_0^gamma (1+z)^3, is best described as a ``de-clustering'', (1+z)^{0.6+/-0.4}. Or equivalently, there is a weak growth of clustering in co-moving co-ordinates, x_0~ (1+z)^{-0.3 +/- 0.2}. This conclusion is supported by the pairwise peculiar velocities which rise slightly, but not significantly, into the past. The Cosmic Virial Theorem applied to the CNOC2 data gives Q Omega_M/b=0.11 +/- 0.04$, where Q is the three point correlation parameter and b the bias. Similarly, galaxy groups have a virial mass-to-light ratio (evolution corrected) of M_{virial}/L_R^{k,e} = 215h Lsun/Msun, or Omega_M=0.15 +/- 0.05., Comment: Presented at Royal Society Discussion Meeting, March 1998, "Large Scale Structure in the Universe" 20 pages, uses rspublic_mod style file included

The CNOC redshift survey of galaxy clusters measures Omega_M from Omega_e(z)= M/L x j/\rho_c which can be applied on a cluster-by-cluster basis. The mass-to-light ratios, M/L, are estimated from rich galaxy clusters, corrected to the field population over the 0.18 to 0.55 redshift range. Since the luminosity density depends on cosmological volumes, the resulting Omega_e(z) has a strong dependence on cosmology which allows us to place the results in the Omega_M-Omega_Lambda plane. The resulting Omega_M declines if Omega_Lambda>0 and we find that Omega_Lambda<1.5., Comment: 4 pages LaTeX. To appear in "Fundamental Parameters in Cosmology," the proceedings of the XXXIIIrd Rencontres de Moriond

The CNOC2 Field Galaxy Redshift Survey presently contains some 5000 galaxy redshifts, plus extensive UBgRI photometry, and is the largest galaxy sample at moderate redshifts 0.1 < z < 0.6. Here we present some preliminary results on the galaxy luminosity function (LF) and its redshift evolution, using a sample of R < 21.5 CNOC2 galaxies, subdivided into early, intermediate, and late types based on their B-R colors relative to non-evolving galaxy models. We find a significant steepening in the faint-end slope alpha of the LF as one proceeds from early to late types. Also, for all galaxy types we find a rate of M* evolution consistent with that from passively evolving galaxy models. Finally, late-type galaxies show positive density evolution with redshift, in contrast to negative or no density evolution for earlier types., Comment: 8 pages, 7 postscript figures, LaTeX, using paspconf.sty. To appear in "The Young Universe", eds. S. D'Odorico, A. Fontana, and E. Giallongo, A.S.P. Conf. Series

The CNOC1 cluster survey measures Omega_M via Oort's method, Omega_M= M/L x j/rho_c, where M/L is the field mass-to-light ratio, j is the field luminosity density and rho_c is the closure density. A wide range of potential systematic effects are explicitly controlled by independently deriving the mean cluster mass profile (finding good agreement with theoretical predictions), the cluster light profile, the redshift evolution of both cluster and field galaxies, the differential evolution between the two, and the field and cluster efficiencies for the conversion of baryons into galaxies. We conclude that Omega_M=0.19+/-0.06 where the errors are objectively evaluated via resampling methods. The redshift evolution of the numbers of clusters per unit co-moving volume over the 0< z < 0.6 range is found to be very slow, as is required for consistency with a low density universe. The evolution of galaxy clustering in the field is compatible with a low density universe, and strongly disfavors models of galaxy evolution that associate low density halos with individual galaxies., Comment: To appear in the 12 th Postdam Cosmology Workshop, "Large Scale Structure: Tracks and Traces", ed. V. Mueller, World Scientific 1998 also available at http://manaslu.astro.utoronto.ca/~carlberg/cnoc/conference/pots.ps.gz

This paper provides a brief description of the CNOC2 Redshift Survey being carried out at CFHT, giving the scope, technique, and current status of the survey, and some preliminary results., Comment: 6 pages, Latex, 4 Postscript figures, To appear in the proceedings for JD 11 (Redshift Surveys in the 21st Century) at the 23rd IAU General Assembly, Kyoto, 1997. Also available (as compressed Postscript) at http://manaslu.astro.utoronto.ca/~carlberg/cnoc/

A redshift sample of 545 field galaxies is used to examine how the galaxy merger rate changes with redshift, and how mergers affect the observed properties of galaxies in close pairs. Close pairs are defined as those with projected separations less than 20 h^{-1} kpc. At a mean redshift of 0.33, it is found that 7.1 +/- 1.4 % of galaxies are in close physical pairs, compared to 4.3 +/- 0.4 % locally. The merger rate is estimated to change with redshift as (1+z)^{2.8 +/- 0.9}. These results are shown to be consistent with previous close pair studies, and provide one of the strongest constraints to date on the redshift dependence of the merger rate. As with earlier studies, no significant differences between the mean properties of paired and isolated galaxies are detected. However, using the subset of confirmed close physical pairs, those which appear to be undergoing interactions or mergers (based on their images) are found to have galaxies with strong emission lines and very blue rest-frame colors. In addition, these pairs have low relative velocities, increasing the likelihood that the galaxies will merge. We interpret this as the first clear evidence of merger-induced star formation occurring in field galaxies at this redshift., Comment: Accepted for publication in the Astrophysical Journal. LaTeX, 37 pages of text, 4 pages of tables (landscape mode), 11 figures. Also available via anonymous ftp (ftp://astroftp.phys.uvic.ca/pub/patton/pairs/)