Out of the 95 primary candidates, 91 have been observed spectroscopically in the framework of the present
survey. Among the 4 unobserved primary candidates (#10, #189, #14 and #26), one is an already known
quasar
(#189 
Q0116-021). Two of the observed primary candidates remain candidates because the
acquired spectra are not of sufficiently good quality to allow definitive identification (#118 and #179).
Therefore, we identified the nature of 90
(89+1) candidates out of 95. Among these 90 objects, we found
Q0111-008.
The U/B excess technique is very good at picking out the majority of the moderately low-redshift
quasars hidden in a field. However, we find symptomatic that the survey that provided the largest absolute
density of such quasars
(Boyle et al. 1987, 1990) was also the one with the lowest efficiency (
38%). Any
attempt to improve the completeness of the survey is bound to a strong loss of efficiency. The only
alternative is to perform a multitechnique survey for
quasars.
The spectra of most of the active galactic nuclei of the present survey are illustrated in Fig. 2 (click here). Although the spectra have been reduced following classical methods for the correction of the atmospheric extinction and for the calibration with standard stars, the flux scale must be considered as totally relative, owing to the various circumstances of the observations such as the seeing, the climatic conditions and so on. We are sorry that a few spectra acquired during A and B runs have not been archived and are thus no longer available.
We collected in Table 6 (click here) all the 140 candidates in order of increasing right ascension. The first two
columns give the right ascension and the declination of the objects for the J2000 equinox (a colon indicates
slightly less precise values). The third column displays the redshift of the identified extragalactic objects (see
Sect. 5.3 for the determination of the redshift). A colon indicates that the redshift is somewhat uncertain
because only one line is present in the acquired spectrum; a question mark
indicates very uncertain values. The fourth column gives either the B1950 name of the quasar
(e.g. Q0107-025A), or an indication on the nature of the object (if it is not a quasar,
e.g. STAR), or simply states that the object is still an unidentified candidate (C). The fifth indicates whether
the object is a primary (P) or a secondary (S) candidate; an asterisk is printed when the spectrum of the
object is available in Fig. 2 (click here). The sixth column contains the internal identification number of the
candidate in the present survey
(e.g. #30 or
NGC 450 # 30; an asterisk refers to a note in Table 7 (click here)) whereas the seventh reports the
identification of the candidate in the NGC 520 survey (e.g.
NGC 450 # 30 
NGC 520 # 173B; see also Surdej et al., in preparation). The three next columns
show the B and U magnitudes of the objects and their 
colour indices (rounded to the nearest
0.05 mag). A colon indicates uncertain values for particular or very faint objects (less than 0.5 mag
above the calibration limits); for these three columns, an over-estimate or an under-estimate is given when a
value could not be measured because lying out of the calibration range. The last column deals with an
identification of the spectroscopic runs as well as with the exposure times in seconds (dual exposures have
been acquired each time). Finding charts for the extragalactic objects are given in the Appendix.
Figure 2: Spectra of the extragalactic objects of the present survey (see also Table 6 and Sect. 5.2); the name
of each object is given. The abscissae represent the wavelength expressed in Å
whereas the ordinates give the flux in arbitrary units
Table 6: The catalogue of quasars and quasar candidates (see Sect. 5.2 for further explanations)
Table 7: Notes to the catalogue of Table 6
Table 8 (click here) contains the observed air wavelengths of the different emission lines for most of the
detected extragalactic objects. The positions have been measured through the fitting of a gaussian onto the
observed profile, and then
corrected for the Doppler effect resulting from the revolution of the Earth around the Sun and from a galactic
rotation velocity, at the distance of the Sun, of 250 km s
. The errors in position are given in the form
of 
1 standard deviation, the latter being essentially representative of the internal consistency of our
measurements. Table 8 (click here) also gives for each object the redshift as derived from the observed lines
(the computations were rigorously performed on the vacuum wavelengths); the final value of z and its
standard deviation are based on those lines seemingly free of any vitiating effect (the lines actually used are
indicated in bold characters). Vertical bars indicate the observed part of the spectrum in the quasar reference
frame. The laboratory wavelength at rest of the Si IV/O IV] blend is taken to be
according to Wills & Netzer (1979). The rest laboratory
wavelength of the C III] intercombination line is taken to be
when the redshift is larger or equal
to 1.8 and 
in the other cases; these values were proposed by Wills
(1980)
and Ferland (1981)
to take into account the possible evolution of the
observed intensity ratio in a blend. As usual, all the wavelengths below
2000 Å are given as vacuum wavelengths. For clarity reasons, Table 8 (click here) has been subdivided into
three parts on the basis of the redshift range. The objects are classified within each part according to
increasing right ascension, as in Table 6 (click here).
A legend explaining the different symbols and additional remarks are given
below the
table. The redshifts from Table 8 (click here) were transferred to the
catalogue of Table 6 (click here) except for the three quasars Q0118-031A, Q0118-031B and
Q0118-031C for which better values are available from Robertson et al.
(1986).
Table 8: Identification table of the emission lines observed in the spectra of the different quasars. The
position of the line and its standard deviation are given in Å. The redshift derived from the line is
mentioned underneath. The last column but one gives the mean redshift attributed to the object and the
related standard deviation based on the line to line dispersion. The bold characters indicate the lines used to
derive the redshift. Vertical lines indicate the observed part of the spectra in the reference frame of the
quasar. The last column refers to additional remarks to be found below the table. The three parts of the table
correspond to different redshift ranges: a) z > 1.5, b) 1.5 > z > 0.5, c) 0.5
> z
Table 9: Total and equivalent widths of the emission lines observed in the spectra of the different quasars.
The configuration of this
table is as that of Table 8. We give, on each first row, the total width at the bottom of the emission lines
(Å) and on each second row, their equivalent width (Å) in the reference frame of the quasar. Again the
three parts correspond to the different redshift ranges
Table 10: Total widths and integrated fluxes of the emission lines observed
in the spectra of objects with a faint continuum. The configuration and legend of this table are as those of
Table 9. We give on each first row the total width at the bottom of the emission lines (Å) and, on the
second row, the emitted flux integrated over the lines (arbitrary units)
For the same objects, we give in Table 9 (click here) the full widths at the bottom of the lines and the equivalent widths in the reference frame of the quasar. Table 9 (click here) is subdivided into three parts as for Table 8 (click here). Finally, let us note that a few objects at low redshift and exhibiting narrow lines over a faint continuum are given separately in Table 10 (click here), along with the full widths at the base and the emitted fluxes integrated over the emission lines.
We present in Figs. 3 (click here), 4 (click here) and 5 (click here) the histograms of the U, B magnitudes
and of the 
indices for the different subsets of candidates.
Figure 3: U magnitude histograms of the objects:
a) primary candidates, b)
primary + secondary candidates; one bin corresponds to 0.2 mag
Figure 4: B magnitude histograms of the objects: a)
primary candidates, b) primary + secondary candidates;
one bin corresponds to 0.2 mag
Figure 5: 
histograms of the objects: a)
quasars and related AGN, b)
primary candidates, c) primary +
secondary candidates; one bin corresponds to 0.2 mag
On the basis of the U-B index (Fig. 5 (click here)), it is clear that the photometric errors are quite large,
as expected: the cut-off that should be located around 
= -0.4 is far from being total and several
objects are redder than 
= -0.2. The cut-off is nevertheless visible around 
= -0.3
for the primary candidates. The exact position of this cut-off is very important to fix the characteristics of the
survey; we thus attempted to refine the above-mentioned values. We generated populations of objects with
tail-like distributions of 
(exponential or linear probability density function, pdf) and a perfect cut-
off. We then added to each object a random error drawn out from a gaussian distribution with zero mean and
= 0.15. For different positions of the cut-off, we computed the two-sample Kolmogorov-Smirnov
statistic which is a measure of the similarity between the generated population and our data. We thus obtain a
mean value of the statistic as well as a dispersion over the simulations, all that as a function of the cut-off
value. For the primary candidates, the mean Kolmogorov-Smirnov statistic exhibits a minimum for a cut-off
of 
= 
. The error represents one standard deviation of the dispersion around the
minimum as deduced from the dispersion of the statistic. For the primary+secondary candidate dataset, we
rather obtain 
= -0.20 which indicates that at least some of the secondary candidates are less
blue than the primary ones, as expected. An alternative approach consists in computing, from the
calibrations, the 
colour index corresponding to strictly equal log(flux) in both images. Of course,
this is not strictly similar to what we actually did because the U image is slightly blurred compared to the
B one; it remains unclear to which aspect the eye is the most sensitive: to the size of the image or to the
central density? Probably to a combination of both but this combination is not necessarily equivalent to the
integrated density and may be variable with the brightness of the object. The 
colour index for
equal log(flux) is clearly a function of the flux. For B between 17
and 19, we have U-B 
-0.3 in good agreement with the above results. Outside that region, the
is slightly bluer (
-0.4).
Let us turn now to the histogram of the B magnitudes. The main cut-off is at B = 19.6: beyond, there is
a marked drop of the bin counts, indicating the limiting magnitude of the survey. Another - more surprising -
- result is the fact that another cut-off seems to be present at B = 19.0. Beyond, there is a marked
decrease of the bin counts where one expects a continuing increase. The origin of this effect is unknown; it
could be either real or induced by an observational problem and deserves further attention. This could also
be related to the above-mentioned possible dependence of the 
cut-off on the brightness of the
object. Our on-going 
survey should shed some light on this problem. The histogram of the U
magnitudes exhibits another behaviour, with the existence of a cut-off at U = 18.8, but with still large
counts beyond. In fact the counts at the faint end vanish progressively with no indication of a limiting
magnitude. The explanation is that, as most of the blue candidates have 
indices between -0.3
and -0.8, the B magnitudes of the bluer objects having U > 18.8 are fainter than the limiting B
magnitude of the survey. These objects are expected to be lost because for some of them only one image
(U) is present and it is mistakenly identified as an isolated B image, typical of the very common red
objects.
We compare our surface density of quasars with the results from other surveys in Table 11 (click here). We also give the power-law suggested by Braccesi et al. (1980; see also Boyle et al. 1987) which beyond B = 19.0 should be considered as an extrapolation. For objects brighter than 18.5, our density compares well with those of other surveys. For the magnitude range 18.5 - 19.0, our values are slightly lower but this can be partly explained by the fact that the secondary candidates are not yet identified; they are particularly numerous in this magnitude range. For the magnitude range 19.0 - 19.5, we have a very strong depletion of quasars indicating that we are far under completeness for this last bin (i.e. the nearest to the limiting magnitude).
Table 11: Surface density of ultraviolet-excess quasars per square degree and per 0.5 magnitude
bin
Out of the 59 quasars originally present in the list of the primary candidates, 6 were already suspected to be
quasars in the framework of the University of Michigan objective-prism survey
(McAlpine & Lewis 1978): 3 probable QSOs UM 314 (z 
2.15), UM 315
(z 
2.09), UM 322 (z 
1.99) and 3 possible ones UM 310 (z 
1.38),
UM 316 (no z) and UM 324 (no z). McAlpine & Feldman (1982) performed the spectroscopic
identification and refined some redshifts:
UM 314 (z = 2.19), UM 315 (z = 2.05), UM 316 (z = 0.96), UM 322 (z = 1.93) and UM 324
(z = 0.35). As for UM 310, it was identified with the Parkes radiosource PKS 0112-017
(Wall et al. 1971) for which Wills & Lynds (1978) proposed a redshift
z = 1.365. We reobserved UM 310 (
Q0112-017),
UM 314 (
Q0115-011), UM 316 (
Q0117-024), UM 322
(
Q0123-021) and UM 324 (
Q0124-021)
and improved the redshift value for the last four quasars. We adopted the published value for UM 315
(
Q0116-021). Later on, Véron-Cetty et al. (1988) redetermined the redshift of UM 324 as
being z = 0.355, a value identical to
the one we derived.
A second Parkes radiosource is present in our field (Ekers 1969):
PKS 0122-00(3)
UM 321 which is a quasar with a redshift z = 1.070
(Lynds 1967).
We did not detect it, as is also the case for a second quasar from the UM survey: UM 320 at z = 2.28
(McAlpine & Feldman 1982).
Finally we did not detect UM 312 which is a possible QSO which seemingly never underwent a
spectroscopic identification.
At the end of 1987, a preliminary version of the data published here was made available via EG's thesis (Gosset 1987a) and also via the fourth version of the Véron-Cetty & Véron catalogue (1989).
Part of the NGC 450 field has been surveyed in the framework of the LBQS survey
(Chaffee et al. 1991). They found, within the common region, 11 strictly LBQS QSOs.
Among these 11 QSOs, 9 were primary candidates in
our survey, one was a tertiary candidate (
= -0.20) and we did not detect the last one.
Chaffee et al. (1991) redetermined the redshifts of our 9 primary candidates; brute
comparative results are given in
Table 12 (click here). One sees that the agreement is rather good
except perhaps for the redshift of Q0109-014. The discrepancy
is most probably due to the effect of the absorption component on the
C IV line (see Table 8 (click here)). It is surprising to notice
that Chaffee et al. (1991) did not detect Q0107-025B (#209),
the twin of Q0107-025A (#210) (see Surdej et al. 1983;
Swings
et al. 1985; Surdej et al. 1986)
whereas they detected the latter one
(see Table 12 (click here)). Chaffee et al. (1991)
also reported the existence of two faint non-LBQS QSOs
Q0107-0031 (
1.753) and Q0108+0030 (
0.428)
that we did not detect.
After inspection of different catalogues and databases, we noticed that
two more quasars present in the NGC 450 field remained undetected; these are MS 0114.3-0123
(V = 19.44, z = 0.559)
discovered as an Einstein Medium Sensitivity Survey (EMSS)
X-ray source (Stocke et al. 1991) and Q0112-014
(B 
20.3, z = 2.20) discovered through a grism or grens survey
in EMSS fields (Anderson & Margon 1987). The latter is too faint for the present
survey; therefore, we derive a
figure of 6 for the number of known quasars missed by our
survey (UM 321 
PKS 0122-00(3), UM 320, MS 0114.3-0123,
Q0110-0009, Q0107-0031 and Q0108+0030).
If their redshifts are potentially accessible for U/B detection
techniques, their exact 
colours remain however unknown.
Table 12: Cross-identification and comparison between the LBQS and the present survey. Odd rows
correspond to LBQS data whereas even ones correspond to our data
