1973ApJ.
.
.184
.
.
.57B
The
Astrophysical
Journal,
184:57-63,
1973
August
15
©
1973.
The
American
Astronomical
Society.
All
rights
reserved.
Printed
in
U.S.A.
A
SEARCH
FOR
HIGH-IONIZATION
REDSHIFT
SYSTEMS
IN
THE
ABSORPTION
SPECTRA
OF
FIVE
QUASARS
J.
N.
Bahcall
and
P.
C.
Joss
Institute
for
Advanced
Study,
Princeton,
New
Jersey
AND
J.
G.
Cohen
Department
of
Astronomy,
University
of
California
at
Berkeley
Received
1973
January
22;
revised
1973
March
26
ABSTRACT
We
have
searched
the
absorption
spectra
of
five
quasars
for
the
presence
of
redshift
systems
dominated
by
the
highly
ionized
doublets
C
iv,
N
v,
and
O
vi,
which
could
be
the
strongest
lines
produced
by
absorbing
clouds
with
collisional
ionization
temperatures
between
10
5
°
and
10
6
°
K.
There
is
at
most
marginal
evidence
for
one
such
system
apiece
in
the
spectra
of
PHL
957
and
4C
05.34,
which
are
the
two
quasars
with
the
largest
known
emission
redshifts.
Highly
ionized
redshift
systems
of
this
type
are
not
widespread
among
the
five
quasars
we
investigated;
the
number
of
redshifts
found
in
the
observed
spectra
is
not
significantly
larger
than
the
number
found
in
similar
random-number
spectra.
Less
than
5
percent
of
the
observed
absorption
lines
are
identified
in
a
statistically
significant
way
by
redshift
systems
of
this
type.
Subject
headings:
quasi-stellar
sources
or
objects
—
redshifts
I.
INTRODUCTION
Systematic
searches
for
redshift
systems
in
the
rich
absorption
spectra
of
quasars
(see
Bahcall
1968;
also
Bahcall
and
Joss
1973,
and
references
therein)
have
been
carried
out
under
the
assumption
that
low-ionization
lines
(especially
La)
will
be
present
if
they
are
in
the
accessible
part
of
the
spectrum;
this
is
equivalent
to
assuming
that
the
absorbing
cloud
is
not
extremely
highly
ionized.
Cohen
(1973)
has
recently
pointed
out,
on
the
basis
of
a
collisional
ionization
model
for
an
absorbing
cloud,
that
the
low-ionization
lines
may
not
be
among
the
strongest
absorption
lines
if
T
^
10
5
°
K,
where
T
is
the
ionization
temperature.
In
fact,
for
a
plausible
abundance
of
heavy
elements
within
the
cloud,
the
strongest
lines
for
10
5
°
^
T
^
10
6
°
K
will
be
due
to
C
iv,
N
v,
and
O
vi
(similar
results
have
been
obtained
by
McKee,
Tarter,
and
Weisheit
1973).
If
redshift
systems
due
to
such
clouds
are
actually
present
in
the
spectra
of
quasars,
they
may
well
have
been
overlooked
in
previous
analyses
by
Bahcall
and
his
collaborators.
In
the
present
paper,
we
present
the
results
of
an
investigation
of
the
rich
absorption
spectra
of
five
quasars
(PKS
0237-23,
TON
1530,
PHL
938,
4C
05.34,
and
PHL
957)
to
determine
whether
high-ionization
redshift
systems
may
be
present.
In
§
II
we
out-
line
our
method
of
analysis,
in
§
III
we
present
our
results
for
the
above
five
quasars,
and
in
§
IV
we
estimate,
with
the
aid
of
random-number
spectra,
the
probabilities
for
apparent
high-ionization
systems
to
arise
by
chance
in
the
spectra
of
the
quasars
we
investigated.
In
the
Appendix,
we
discuss
in
more
detail
the
method
of
construction
of
the
random-number
spectra.
II.
METHOD
We
used
the
method
of
analysis
of
rich
absorption-line
spectra
developed
by
Bahcall
(1968).
We
included
16
lines
in
our
list
of
standard
lines:
O
vi
ÀÀ1031.95,
1037.63;
57
©
American
Astronomical
Society
•
Provided
by
the
NASA
Astrophysics
Data
System
1973ApJ.
.
.184
.
.
.57B
58
J.
BAHCALL,
P.
JOSS,
AND
J.
COHEN
Vol.
184
N
v
ÀÀ1238.81,
1242.80;
C
iv
AÀ1548.20,
1550.77;
C
iv
(av)
À1549.50;
Si
iv
ÀÀ1393.76,
1402.77;
C
m
À977.03;
Fe
m
Al
122.53;
Si
m
A1206.51;
Al
m
AA1854.72,
1862.78;
La;
and
L/3.
We
required
that
the
absolute
wavelength
discrepancy
between
an
observed
line
and
a
redshifted
standard
line
be
less
than
or
equal
to
2
Á
for
an
acceptable
identification.
In
order
to
search
for
high-ionization
redshift
systems,
we
established
a
set
of
formal
rules
to
determine
whether
a
candidate
system
would
be
accepted.
The
rules
were
:
1.
At
least
two
complete
doublets
among
the
three
doublets
(C
iv,
N
v,
O
vi)
must
be
present.
2.
The
ratio
of
line
strengths
for
each
accepted
doublet
must
be
consistent
with
atomic
physics
and
spectroscopic
limitations.
3.
The
two
accepted
doublets
must
contain
at
least
two
lines
of
strength
greater
than
or
equal
to
2.
By
trial
and
error,
we
found
that
less
stringent
rules
resulted
in
an
undesirably
large
number
of
acceptable
systems
within
random-number
spectra
having
the
same
essential
characteristics
as
the
observed
spectra
(see
§
IV
and
the
Appendix).
The
same
difficulty
was
encountered
when
we
attempted
to
include
in
our
analysis
the
absorption
lines
in
PHL
957
observed
by
Lowrance
et
al.
(1972)
with
a
high-resolution
integrating
television
system.
In
other
respects,
our
method
of
search
was
identical
to
that
of
previous
analyses
(see
Bahcall
and
Joss
1973,
and
references
therein).
The
observational
data
were
taken
from
Bahcall,
Greenstein,
and
Sargent
(1968)
(for
PKS
0237
—
23),
Bahcall,
Osmer,
and
Schmidt
(1969)
(for
TON
1530),
Burbidge,
Lynds,
and
Stockton
(1968)
(for
PHL
938),
Lynds
(1971)
(for
4C
05.34),
and
Lowrance
et
al.
(1972)
(for
PHL
957).
III.
RESULTS
FOR
FIVE
QUASARS
We
have
searched
the
observed
spectra
of
five
quasars
for
acceptable
high-ionization
redshift
systems
between
z
=
4.0
and
a
minimum
redshift
z
min
,
the
value
of
which
TABLE
1
High-Ionization
Absorption
Redshift
Systems
in
Two
Quasars*
Observed
Wavelength
Identifications
in
Wavelength
Discrepancy
Low-Ionization
(Â)
Strength
Identification*}*
(Â)
Systems
4C
05.34:
emission
redshift
z
erri
=
2.877;
absorption
redshift
z
a
bs
=
2.5711
3683.54
3
O
vi
(2)
1031.95
+1.7
Lj3
1025.72
(z
=
2.5925)
fLy
972.54
(z
=
2.8106)
3706.10
3D
O
vi
(1)
1037.63
-0.6
^
O
vi
1031.95
(z
=
2.5925)
lC
ii
1334.53
(z
=
1.7758)
[
4010
-
4
8
1
Fe
m
1122.53
-1.8
{|
HO^z
=
S)]
4422.83
4B
N
v
(2)
1238.81
+1.1
None
4436.74
3
N
v
(1)
1242.80
+1.4
Si
iv
1393.76
(z
=
2.1819)
PHL
957:
z
em
=
2.69;
z
abs
=
2.6181
3735.0.......
2W
O
vi
(2)
1031.95
-1.3
None
3753.9
2W
O
vi
(1)
1037.63
+0.4
LjS
1025.72
(z
=
2.6613)
4482.5
1
Nv
(2)
1238.81
-0.4
None
4495.0
1
N
v
(1)
1242.80
+1.6
Si
iv
1393.76
(z
=
2.2250)
*
Letters
following
the
line
strengths
stand
for
diffuse
(D),
broad
(B),
and
wide
(W),
as
denoted
by
the
observers.
The
identification
in
brackets
is
questionable
and
was
not
considered
as
evidence
supporting
the
other
identifications.
t
Relative
intrinsic
strengths
between
the
members
of
each
doublet
are
given
in
parentheses.
©
American
Astronomical
Society
•
Provided
by
the
NASA
Astrophysics
Data
System
1973ApJ.
.
.184
.
.
.57B
No.
1,
1973
SEARCH
FOR
HIGH-IONIZATION
REDSHIFTS
59
was
set
equal
to
the
minimum
redshift
for
which
two
of
the
three
doublets
C
iv,
N
v,
and
O
vi
fall
within
the
observationally
accessible
portion
of
the
spectrum.
The
value
of
¿min
varies
from
~
1.5
for
TON
1530
to
~
1.8
for
4C
05.34.
We
found
one
acceptable
system
apiece
in
the
spectra
of
PHL
957
and
4C
05.34
at
redshifts
of
z
=
2.6181
and
z
=
2.5711,
respectively;
the
properties
of
these
systems
are
summarized
in
table
1.
The
acceptability
of
both
systems
relies
upon
the
identification
of
the
doublets
of
N
v
and
O
vi.
Another
system
(z
=
2.2055)
in
the
spectrum
of
PHL
957
obeys
all
of
our
formal
rules,
but
we
rejected
it
because
of
an
unacceptably
large
wavelength
dis-
crepancy
(3.6
Â)
between
the
two
lines
of
the
closely
spaced
doublet
of
O
vi.
(If
this
doublet
is
discounted,
the
system
is
unacceptable.)
This
system
is
virtually
identical
to
the
system
z
=
2.2056
in
table
5
of
Lowrance
et
al
(1972).
We
found
no
acceptable
systems
in
the
spectra
of
three
other
quasars
(PKS
0237
—
23,
TON
1530,
and
PHL
938).
We
also
note
that,
among
the
eight
identified
lines
in
the
two
acceptable
systems,
five
have
already
been
identified
in
previous
searches
for
low-ionization
systems.
Moreover,
among
these
five
previously
identified
lines,
four
were
essential
to
the
acceptability
of
low-ionization
systems
in
which
they
were
identified.
IV.
RANDOM-NUMBER
SPECTRA
In
order
to
estimate
the
probability
that
an
acceptable
high-ionization
redshift
system
will
occur
by
chance,
we
have
generated
and
analyzed
20
random-number
spectra
corresponding
to
each
quasar.
These
spectra
were
required
to
have
the
same
essential
characteristics
as
the
observed
spectra,
e.g.,
the
minimum
separations
between
lines
were
the
same
as
those
imposed
on
the
observed
spectra
by
the
spectro-
scopic
techniques
and
observational
conditions.
Further
details
concerning
the
char-
acteristics
of
the
random-number
spectra
are
given
by
Bahcall
(1968)
(for
PKS
0237-23),
Bahcall
et
al.
(1969)
(for
TON
1530),
Bahcall
and
Feldman
(1970)
(for
PHL
938),
Bahcall
and
Goldsmith
(1971)
(for
4C
05.34),
and
Bahcall
and
Joss
(1973)
(for
PHL
957).
(See
also
the
Appendix
to
the
present
paper.)
In
all
cases,
we
checked
to
ensure
that
acceptable
redshift
systems
in
these
spectra
would
not
have
been
accepted
in
previous
searches
for
low-ionization
systems.
The
results
of
our
analysis
of
the
random-number
spectra
are
summarized
in
table
2.
On
the
basis
of
the
number
of
acceptable
systems
in
the
random-number
spectra,
we
estimate
that
the
probability
of
the
two
acceptable
systems
having
occurred
by
chance
in
the
observed
spectra
of
PHL
957
and
4C
05.34
is
~
10
percent,
and
that
the
prob-
ability
of
two
acceptable
systems
having
occurred
by
chance
among
all
five
quasars
is
~25
percent.
We
note
that
five
out
of
the
14
acceptable
systems
in
the
random-
number
spectra
have
redshifts
appreciably
larger
than
the
emission
redshifts
of
the
respective
quasars,
while
no
acceptable
absorption
redshift
much
in
excess
of
the
emission
redshift
has
yet
been
found
in
any
observed
quasar
spectrum.
However,
this
consideration
is
of
limited
value
in
the
context
of
the
present
probability
estimates,
because
high-ionization
redshift
systems
have
also
not
yet
been
reported
in
the
absorption
spectra
of
quasars.
The
total
number
of
lines
in
the
quasar
absorption
spectra
we
have
studied
is
267.
Since
the
two
candidates
for
high-ionization
absorption
redshift
systems
explain
at
most
nine
lines,
we
conclude
that
the
percentage
of
absorption
lines
which
can
be
explained
with
statistical
validity
as
having
been
formed
under
the
conditions
of
high
collisional
ionization
considered
by
Cohen
(1973)
and
McKee
et
al.
(1973)
is
less
than
5
percent.
Another
way
of
summarizing
this
conclusion
is
to
note
that
a
total
of
23
redshifts
have
been
previously
identified
in
these
quasars
using
the
same
techniques
employed
in
the
present
paper.
Thus,
at
most
2/25
of
the
redshifts
identified
in
this
statistically
significant
way
are
likely
to
have
been
produced
under
conditions
of
high
collisional
ionization.
©
American
Astronomical
Society
•
Provided
by
the
NASA
Astrophysics
Data
System
1973ApJ.
.
.184
.
.
.57B
60
J.
BAHCALL,
P.
JOSS,
AND
J.
COHEN
TABLE
2
Acceptable
Redshifts
among
the
Random-Number
Spectra
Number
of
Acceptable
Acceptable
Redshifts
in
Redshifts
per
Random-
Object
20
Random-Number
Spectra*
Number
Spectrumt
PHL
957
2.55
0.20
±
0.51
2.49
2.31Î
1.96Î
4C
05.34
3.48
0.15
±
0.36
2.91
2.28
PHL
938
2.69
0.20
±
0.40
2.55
1.79
1.72
TON
1530
None
0
PKS
0237-23
2.76
0.15
±
0.36
1.79
1.70
*
Italicized
redshifts
are
larger
than
the
observed
emission
redshift
of
the
quasar,
t
The
quoted
errors
are
one
standard
deviation
per
random-number
spectrum.
t
These
two
redshifts
were
in
the
same
random-number
spectrum.
This
is
the
only
case
where
more
than
one
acceptable
redshift
was
found
in
a
single
spectrum.
This
work
was
initiated
while
one
of
us
(J.
N.
B.)
was
a
visiting
professor
in
the
astronomy
department
of
the
University
of
California
at
Berkeley;
J.
N.
B.
wishes
to
express
his
gratitude
for
the
hospitality
and
stimulation
provided
by
the
members
of
the
Berkeley
astronomy
department.
This
work
was
supported
in
part
by
NSF
grant
16147
A#1
and
the
Miller
Institute
for
Basic
Research.
APPENDIX
In
this
Appendix,
we
discuss
the
method
by
which
the
random-number
spectra
are
generated.
The
purpose
of
the
random-number
spectra
is
to
determine
the
statistical
validity
of
the
candidate
redshifts
that
are
found
in
the
observed
spectra.
Hélice,
the
random-
number
spectra
are
generated
in
a
way
that
simulates
the
limitations
imposed
on
the
observed
spectra
by
the
spectroscopic
techniques
and
observational
conditions;
the
random-number
spectra
are
thus
superficially
indistinguishable
from
the
real
spectra.
In
particular,
the
minimum
and
maximum
wavelengths
(A
mln
and
A
max
,
respectively)
allowed
for
the'lines
in
a
random-number
spectrum
are
set
equal
to
the
minimum
and
maximum
wavelengths
accessible
to
the
observed
spectrum;
the
minimum
separation
A
Amin
between
each
pair
of
lines
in
a
random-number
spectrum
is
set
equal
to
the
spectral
resolution
of
the
observed
spectrum;
the
total
number
n
of
lines
in
each
random-number
spectrum
is
set
equal
to
the
total
number
of
observed
lines;
and
the
distribution
of
line
strengths
and
other
line
characteristics
is
set
equal
to
the
observed
distribution
(e.g.,
if
there
are
two
wide
lines
of
strength
3
in
the
observed
spectrum,
then
two
randomly
chosen
lines
in
each
random-number
spectrum
are
considered
to
be
wide
and
of
strength
3).
(See
Bahcall
1968
and
Bahcall
and
Joss
1973
for
additional
details
concerning
the
generation
of
the
random-number
spectra.)
©
American
Astronomical
Society
•
Provided
by
the
NASA
Astrophysics
Data
System
1973ApJ.
.
.184
.
.
.57B
TABLE
A-l
A
comparison
of
the
observed
ab
sorption-line
spectrum
of
4C
05.
34
with
two
typical
random-number
spectra
(X.
.
=
3500&.,
X
=6000$.,
AX
.
=6$,
n
=
93).
mm
max
mm
Observed
Spectrum
Wavelength(X)
Strength^
Random-number
Spectrum
#
1
Wavelength($)
Strength
1
Random
-
numb
e
r
Spectrum
#
2
Wavelength($)
Strength'
1
3497.85
3514.81
3535.98
3565.61
3582.
46
3592.
18
3603.32
3630.26
3641.
86
3668.39
3683.54
3706.10
3727.17
3744.74
3766.27
3779.43
3816.
09
3843.96
3867.06
3874.82
3882.
94
3894.82
3908.29
3965.40
3974.34
3985.83
3997.07
4010.
48
4019.67
4042.
23
4053.04
4061.
50
4069.19
4084.42
4142.
34
4154.96
4163.
15
4171.
36
4185.46
4197.38
4215.43
4223.
59
4233.
72
4246.
98
4256.
62
4296.
49
4304.62
2
2
2
3D
3
0
3
0
3
2
3
3D
4B
1
3
2
2D
1
3D
0
1
2
ID
1
1
2
5
1
3
1
3
1
3
ID
0
0
1
1
2D
2
2
4
0
3
2
3
1
3523.
30
3591.
19
3602.
92
3705.78
3763.98
3792.
08
3798.91
3835.99
3869.55
3914.35
3924.81
3937.94
3956.61
3988.19
4023.70
4076.32
4111.
23
4152.
67
4184.18
4201.07
4243.88
4256.
00
4266.18
4324.60
4338.26
4352.
18
4358.24
4365.21
4418.
41
4432.
13
4446.15
4452.
67
4519.05
4532.
05
4621.
73
4656.89
4662.
96
4684.57
4733.20
4806.53
4852.
62
4861.
75
4880.21
4900.10
4909.73
4924.22
4934.06
0
1
0D
1
3
1
0D
3
0
3
1
ID
1
3
3D
ID
0
3
3
3
1
1
1
2
2
0
2
ID
2
0
5
1
3
5
1
3
1
4
1
4
2
3
3
3
2D
0
0
3561.
24
3572.
72
3599.
68
3611.45
3627.
59
3643.24
3659.23
3692.
97
3709.26
3747.
34
3782.
47
3793.04
3823.
24
3861.
85
3901.
81
3939.39
3958.
61
4018.
55
4037.78
4047.19
4056.02
4070.67
4133.38
4166.56
4213.
48
4249.75
4259.
82
4284.93
4297.
61
4313.47
4328.88
4338.
91
4348.87
4375.27
4388.
52
4418.
08
4454.23
4461.
34
4503.57
4519.
52
4528.
46
4573.64
4595.82
4613.70
4620.
88.
4635.74
4642.
92
2
0
4B
4
1
1
2
2
4B
2
1
1
3
1
2
0
3D
2
0
1
3
1
ID
1
3D
4
2
3
1
2
3
1
0D
3
1
3
1
0
2
0
3
5
5
0
2
3
4
*
Letters
following
line
strength
indicate
"broad"
(B)
and
"diffuse"
(D),
as
denoted
by
Lynds
(1971).
©
American
Astronomical
Society
•
Provided
by
the
NASA
Astrophysics
Data
System