Abstract
This thesis is concerned with the cosmological evolution of quasars. I describe the construction of the Edinburgh Multicolour Survey from COSMOS scans of UK Schmidt
photographic plates, taken in U BVRI wavebands at high galactic latitude in a contiguous area of 0.1 steradians (13 UK Schmidt fields). Two plates are taken close together
in time in each waveband in each field so that spurious detections can be eliminated,
and the errors on the measured magnitudes reduced. The raw COSMOS datasets were
calibrated using photoelectric and CCD sequences in each waveband in each field. Systematic errors in the calibration due to “field effects” (variations in image size across
each plate) are minimised by using the colours of the stars on each plate. Differences
between the plates in each waveband are minimised. Differences in each waveband between fields are minimised by using the spatial distribution of stars in the survey, and
requiring it to be uniform across the whole survey area. The calibration of the “ worst”
(as judged by the level of field effects) is tied in with that of the “best” plates. The
final dataset is uniformly and accurately calibrated across the entire survey area. The
systematic error in the COSMOS-measured magnitudes at B = 15 — 16 is O.Olra. The
mis error at B = 17 — 18 (where most of the quasars are) is 0.09m.
The Edinburgh Multicolour Survey was used to select a sample of bright UVX candidates.
I then describe how follow-up spectroscopy was carried out to determine the nature of
the candidates, and in the case of the quasars, to measure their redshifts. I compare the
surface density of quasars found in this way to that measured by previous surveys, in
particular the Palomar-Green Survey (Schmidt & Green 1983). Until now, the PalomarGreen Survey was the principal source of bright, optically selected quasars and, as such,
has been used to determine the nature of quasar evolution. However I show in Chapter 4
that the Palomar-Green Survey is significantly incomplete, by a factor 3, when compared
to the Edinburgh Survey, and that the slope of the differential log(number)-magnitude
relation is 0.73 ± 0.07, substantially less than previous determinations. I then use the
Edinburgh Survey, together with the fainter, larger AAT survey (Boyle et al. 1990) to
calculate the luminosity function in redshift slices. I compare this luminosity function
with the accepted model; Pure Luminosity Evolution (Boyle 1991), in two ways; by
comparing the model and observed cumulative distributions in absolute magnitude, and
by fitting a function to the observed differential luminosity function. I show that Pure
Luminosity Evolution is not a good description of the data, and it appears that the
luminous part of the luminosity function changes shape as a function of redshift, such
that the most luminous quasars evolve the least. I discuss the implications of this result,
in particular with respect to hierarchical galaxy formation models and with respect to
quasar lifetimes. When Pure Luminosity Evolution was the best-fit model and there was
no need for density evolution, workers favoured long lifetimes. I argue that there is no
justification for that, and that nothing can be known about quasar lifetimes by studying
the luminosity function. I discuss possible future work.