Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.mrao.cam.ac.uk/~anthony/capetown_paper/lasenby.ps Äàòà èçìåíåíèÿ: Wed May 1 21:25:22 2002 Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 07:23:07 2012 Êîäèðîâêà: Ïîèñêîâûå ñëîâà: 2df survey

CMB anisotropies: recent measurements and
interpretation
Anthony Lasenbyy
Astrophysics Group,
Cavendish Laboratory,
Madingley Road,
Cambridge, CB3 0HE, U.K.
Abstract. The aim of this contribution is to give a brief survey of the current
state of cosmic microwave background experiments and results. The eects of
systematics and experimental uncertainties are emphasized, and a summary given
of our current knowledge of cosmological parameters, using the latest data. Some
future experiments are also discussed.
1. Introduction
The cosmic microwave background (CMB) is currently providing us with
extraordinarily detailed and useful information about the early universe, and about
the parameters which control the fate of the universe. It is quite certainly an exciting
time in CMB studies, with many new experiments contributing to the rush of data,
and with the promise of even more accurate and detailed information to come. This
contribution seeks to give a summary of these recent experiments, and will draw
particular attention to the eects of remaining uncertainties and likely systematics
in the data. It is vital to understand these in order to have a proper appreciation
of the signicance of the constraints and the results which the data provide. To this
end, we give a brief summary of experimental methods and contaminants, and then
a description of the balloon and interferometer experiments which are currently in
progress. The implications of these results in terms of parameter estimation are then
discussed, and we close with a summary of new and planned experiments, assessing
their signicance in terms of dierent regions of parameter and model space.
Most interest in CMB studies is currently focussed on the total intensity CMB
power spectrum. In the (near) future, results from polarization studies will become of
great importance and actual features on the maps will become of increasing signicance
in connection with studies of non-Gaussianity. The total intensity CMB power
spectrum is dened and discussed in other papers in this volume, so here we highlight
only the most important features. In Fig. 1 we show a typical power spectrum with its
dependence on major parameters indicated. The power spectrum is dened in terms
of a decomposition of the sky into spherical harmonics:
ôT
T
(; ) =
X
`;m
a`mY`m (; );
y E-mail address: a.n.lasenby@mrao.cam.ac.uk

CMB anisotropies: recent measurements and interpretation 2
Figure 1. Schematic form of expected CMB power spectrum, showing important
dependencies.
The rotationally invariant power spectrum is dened from this by
C ` = hja `m j 2 i:
What is plotted in Fig. 1 is `(` + 1)C ` , which is the power per unit log interval in `.
A rough way of translating between ` and angular scale is indicated in the gure.
Physical lengths at recombination get translated to an angle on the sky via the
angular diameter distance formula. This is mainly a function
of
total so the left/right
position of the peaks in the power spectrum as a function of ` is a sensitive indicator of
the total energy density of the universe (baryonic, dark and an eective cosmological
constant component). This is shown as the double arrow over the rst peak in the
spectrum: a universe above critical density has its peaks shifted to the left relative
to this diagram, while a universe below critical density has its peaks shifted to the
right. The spacing and height of the secondary peaks in the power spectrum depend
on the detailed physics which occurs during the recombination epoch, and this is
indicated by the dependence upon the baryon
density
b , the Hubble constant H 0 and
the sound speed at that epoch. There is a roughly exponential cuto in primordial
perturbations expected as one goes to higher ` (smaller angular scale) dictated by the
combined eects of the total optical depth,  , between recombination and the present,
and processes such as Silk damping during recombination.
Two other important parameters concern the initial input power spectrum coming
from in ation. The parameter n, the slope of the primordial power-law power
spectrum is dened via
hjô k j 2 i / k n

CMB anisotropies: recent measurements and interpretation 3
where the ô k s are the initial matter power spectrum. In ation (in its most common
forms) predicts n  1. This is what would give a roughly at slope in the CMB power
spectrum on large scales, and corresponds to the region labelled `SW plateau' in the
diagram, named after the Sachs-Wolfe eect [1], which dominates in this region.
The second is the ratio r of tensor (gravitational wave) to scalar (density)
perturbations. This is often assumed to be small, but in various versions of
in ation could be signicant. (Note in ation predicts very small vector (vorticity)
perturbations.)
So, relative to a particular model, one can do detailed tting of the major
parameters(
m
,
b , H 0 , n etc.). The CMB will provide ways to measure these with
unprecedented accuracy, and one can simultaneously check the details of in ationary
predictions and the particle physics potential presumed to give rise to in ation.
2. Experimental Problems and Solutions
The detection of CMB anisotropy at the level
T=T  10 5 is a challenging
problem and a wide range of experimental diôculties occur when conceiving and
building an experiment. Contaminating foregrounds present an obvious problem.
The anisotropic components that are of essential interest are (i) The Galactic dust
emission which becomes signicant at high frequencies (typically > 100 GHz); (ii)
The Galactic thermal (free-free) emission and non-thermal (synchrotron) radiation
which are signicant at frequencies lower than typically  30 GHz; (iii) The presence
of point-like discrete sources; (iv) Atmospheric emission, which is the dominating
source of contamination for ground- and balloon-based experiments, in particular at
frequencies higher than  10 GHz; (v) Finally a possibility has emerged recently for
a `spinning dust' contribution in the range 10{100 GHz.
Concerning point (iv) above, three basic techniques, which are all still being used,
have been developed in order to ght against the atmospheric emission problem:
The Tenerife experiments (e.g. [2]) were one of the pioneers of the switched beam
method. In this case the telescope switches rapidly between two or more beams so
that a dierential measurement can be made between two dierent patches of the sky,
allowing one to lter out the atmospheric variations.
A more recent version of the switched beam method is the scanned beam method,
used very successfully by e.g. by the Saskatoon [3] and Python telescopes. These
systems have a single receiver in front of which a continuously moving mirror allows
scanning of dierent patches of the sky. An optimum pattern of motion by the mirror
can then be re-synthesised by software. This technique provides a great exibility
regarding the angular-scale of the observations and the Saskatoon telescope was very
successful in using this system to provide results on a range of angular scales. Most
balloon observations are also made using a scanning technique.
Finally, an alternative to dierential measurement is the use of interferometric
techniques. Here, the output signals from each of the baseline horns are cross-
correlated so that the Fourier coeôcients of the sky are measured. In this fashion
one can very eôciently remove the atmospheric component in order to reconstruct a
cleaned temperature map of the CMB. This has become the method of choice for most
ground-based experiments now, since it eliminates the atmosphere most eectively out
of all the methods. However, so far it has been used solely with heterodyne receivers
rather than the bolometric receivers used on balloons, and so tends to need a much

CMB anisotropies: recent measurements and interpretation 4
longer integration time to reach a given sensitivity | this is not such a problem from
the ground of course, and can be partially oset if large bandwidths are used.
As regards the other contaminants mentioned above, a natural solution is to run
the experiment at a suitable frequency so that the contaminants are kept low. There
exists a window between  10 and  40 GHz where both ground-level atmospheric
uctuations and Galactic emissions should be lower than the typical CMB anisotropies,
and many experiments have occupied this spectral range. However, in order to reach
the level of accuracy needed, spectral discrimination of foregrounds using multi-
frequency data is highly desirable. This takes the form of either widely spaced
frequencies giving a good `lever-arm' in spectral discrimination (this has been adopted
by most balloon experiments), or a closely spaced set of frequencies which allows
good accuracy in subtraction of a particular known component (as used e.g. in
several ground experiments). For more details both of the properties of the dierent
foreground components and of the methods used to combat them, see e.g. [4]. It
remains a perhaps remarkable fact, however, that for most of the current CMB data,
not very much correction for Galactic contamination has been necessary over the
whole range of primary CMB frequencies from about 30 GHz to 200 GHz. In fact,
of more importance in its impact so far have been systematic eects associated with
lack of knowledge of various parameters of the instruments, such as overall calibration,
beamwidth and pointing uncertainties. To discuss these, we now turn to a detailed
examination of each of several of the major experiments which have given us new and
exciting results over the last couple of years.
3. BOOMERANG
BOOMERANG is a scanning balloon experiment designed to cover the ` range
50  < `  < 1000. It uses the technique of `long duration ballooning'. Here, the
experimental package is launched into winds which will carry it either a long distance
in roughly a straight line before recovery (this is the plan for the ARCHEOPS
experiment), or, as in the case of BOOMERANG, around a large circle back to near
the launch point. The circumpolar winds at Antarctica provide a way of achieving
the latter. The rst long duration balloon ight of this kind was in 1993 (a solar
physics ight). The BOOMERANG team have so far carried out one long duration
ight (December 1998/January 1999), now called the Antarctica or LDB ight, with
another scheduled for December 2002/January 2003. This future ight will include a
polarization capability. There was also a `test ight' on North America in 1997, from
which the rst power spectrum results were released [5]. Each ight had a dierent
conguration for the instrument, with a quite dierent amount of data taken.
The focal plane arrangement for the North American test ight had two
frequencies with 6 bolometers in total. 4.5 hours of data on the CMB were achieved,
with approximately 600 K sensitivity in 4,000 16 arcminute pixels. The 150 GHz
channel was the one mainly used for power spectrum estimation. A brute-force
likelihood approach was carried out in 23,561 one-third beam-sized 6:9 0 pixels, with
the results given in [5, 6].
For the Antarctica ight [7, 8], coverage of 4 frequencies with 16 bolometers in
total were available. The ight was basically perfect rst time, and the balloon gondala
(the scientic package) was picked up just 20 miles from the original launch point
after circling the South Pole. 190 hours of primordial CMB data on an approximately
30 ô 40 ô region were obtained and about 18 hours of data on three Sunyaev-Zeldovich

CMB anisotropies: recent measurements and interpretation 5
100 200 300 400 500 600 700 800
0
10
20
30
40
50
60
70
80
90
Inverse angular scale p /q -1
rms
temperature
fluctuation
DT l
(µK)
MAXIMA-1
BOOMERANG/1%
BOOMERANG/NA
COBE
h=0.65, W m=0.3, W L=0.7
Fit to Boomerang/1%
SLB 09-May-2000
Figure 2. A linear scale comparison of the BOOMERANG North America (NA)
and Antarctica (1%) results with MAXIMA-1. Also shown is a `standard' at
model and a t to the BOOMERANG data alone. Note the `Inverse angular
scale' can be read directly as `.
clusters.
It is fair to say that the maps obtained [7] represented the start of a qualitatively
new age in CMB astronomy. For the rst time, primordial uctuations were clearly
visible over a large area and at several (here 3) frequencies, in a way which removed
any lingering doubt as to their CMB origin. Due to pointing problems (see below),
which limited the eective resolution, the initial analysis was limited to just 8000 14-
arcminute pixels in just one of the 150 GHz detectors. This provided results up to an
` of 600, although the instrument intrinsically has the sensitivity to reach `  800.
The initial power spectrum results are shown in Fig. 2, in comparison with the
initial MAXIMA-1 results (see below). They were quite surprising, in that although a
at (perhaps even closed!) universe was conrmed, the second Doppler peak appears
somewhat suppressed, if it is present at all. Much eort was then spent on seeing
the implications of both this and the MAXIMA data (e.g. [8, 9]), with the most
likely outcome being the need for a
higher
baryon h 2 than usually derived from
nucleosynthesis constraints, and perhaps an overall slope in the primordial power
spectrum from in ation (n dierent from 1). Since a
higher
baryon h 2 tends to suppress
the second peak while raising the third peak, this placed great signicance on the
question of the height of the third peak, when/if it was detected. Theoretical work was
also carried out on whether the data could be explained by more exotic possibilities,
such as variation of the ne structure constant with time [10, 11]. This was stimulated
both by the perceived problem with the second peak, together with the fact that the
BOOMERANG data appeared to favour a marginally closed universe, since it gave
a peak position of ` peak = 197  6 as against the value of about 220 expected for a
at universe. A smaller ne structure constant  at recombination can explain both
of these by delaying the epoch of last scattering. These points are mentioned here

CMB anisotropies: recent measurements and interpretation 6
Figure 3. The BOOMERANG data used in the new analysis of Nettereld et al
[12]. The three circled regions mark the positions of known point sources and the
ellipse shows the part of the CMB map actually used.
as an illustration of both the signicance of the BOOMERANG results, but also as
a warning of the dangers of over-interpretation, and of how experimental results can
change with time before they are fully settled down.
The original Boomerang results (B00) came out in Bernardis et al [7] in April
2000, and then about a year later a new analysis of the data appeared in Nettereld et
al [12]. This was based on an increased area of sky, (1.8% of the entire sky, as against
1% in the earlier results) and on the data from four 150 GHz detectors as against
eectively only one. Since data was used from the entire ight (as against just the
second half only) this gave about 17 times more data than for the previous analysis.
The new sky area and the CMB map used are shown in Fig. 3.
The key to the most signicant changes in the results, however, comes from
understanding the dual diôculties of pointing and calibration uncertainty. The planets
usually used in balloon experiments for beam maps and calibration are low in the sky
at the South Pole. In addition, working in the Austral summer means that stars
are not available for tracking and pointing | the Sun is the main object which has
to be used in real time, along with Galactic HII regions in the subsequent stages of
analysis. Because of these diôculties, it is now understood that the pointing solution
used in B00 produced an eective beam size of 12:7  2:0 arcmin. However, the
beam size assumed in the analysis was 10:0  1:0 arcmin. This means the power

CMB anisotropies: recent measurements and interpretation 7
Figure 4. The new BOOMERANG results from Nettereld et al [12]. See text
for explanation.
spectrum was systematically underestimated at small angular scales (high `). The
remaining uncertainty in eective beam size is still 13%, and this has progressively
larger eects on both possible systematic and noise error as ` increases. In addition,
a new calibration solution implied an overall increase in gain of about 20%, uniform
in C ` , relative to previous values.
The combined eect of these changes can be seen in the new results, shown
in Fig. 4. This gure is the central result of BOOMERANG to date and is worth
explaining in detail. First of all, what one is seeing is the results of two independent
analyses carried out in dierent bins in ` space. Thus the points with central squares
(blue in colour) form one set and those with central triangles (red) form another.
The two sets are from point to point within themselves nearly uncorrelated, but
between neighbouring points taken from alternate sets there is a high correlation
(approx 50%), due to the binnings not being independent. This has now become the
norm for displaying results in `-space, so is worth understanding in relation to worries
about (e.g.) apparently low scatter versus the errors.
Secondly, there is a remaining overall calibration uncertainty of about 20% in the
units of this plot. This is large enough that it needs to be properly taken into account
in any analysis, e.g. by analytic marginalisation over the uncertainty (see e.g. Bridle

CMB anisotropies: recent measurements and interpretation 8
Figure 5. Likelihood contours in
the
b h 2 versus Ns plane using BOOMERANG
and COBE data alone (from [14]). The vertical lines show the standard
nucleosynthesis constraints.
et al [13]). To this error, aecting each point equally, should be added the results
of remaining lack of knowledge in the pointing solution and beamsize uncertainties.
The possible systematic eects of these are illustrated by the black triangles. These
show the 1 range still remaining from these | all the points move systematically
either up or down according as the eective beamwidth has been underestimated or
overestimated (respectively). Again, it is possible to take these uncertainties into
account analytically [13].
With these caveats, we can now discuss the signicant features of the above power
spectrum. Unlike with the earlier BOOMERANG analysis [7], there is apparently now
good evidence for peak 2 (and possibly peak 3) in the power spectrum. The evidence
for secondary peaks here has to be taken with caution | in particular the visual
impression is complicated by the `double binning' which means that successive points
are highly correlated, as explained above. (The formal signicance level for peaks
after the rst is probably about 2 | see Bernardis et al [14].) However it is certainly
the case that ts to the power spectrum no longer call for a value
of
b h 2 which
con icts with standard nucleosynthesis. It is now found (in Nettereld et al [12])
that
b h 2 = 0:022 +0:004
0:003 , whereas (with the old
analysis)
b h 2 = 0:036 +0:005
0:005 was
obtained. The current best standard nucleosynthesis range (e.g. Burles et al [15]) is

b h 2 = 0:020 +0:002
0:002 (95% condence). This is shown visually in Fig. 5. The presence

CMB anisotropies: recent measurements and interpretation 9
of data out to `  1000 is particularly signicant in helping break
the
b h 2 versus n s
degeneracy. In particular, the third peak behaves dierently under variations in these
two quantities as against the second and rst peaks, thus rening the signicance
of the estimates possible. Note, however, that the results from the `BOOMERANG
alone' analyses presented in [12] and [14] are generated without including a tensor
contribution. This means that they are perhaps overly optimistic as compared to
other analyses (see below).
Two other signicant changes have occurred in the new BOOMERANG results
as against the old. Firstly, the rst peak now occurs closer to ` = 220, meaning that
any evidence for a closed model has faded. Secondly, due to the overall revision of
the calibration upwards, the tendency of the previous results to give high values of
the Hubble constant (other parameters being xed at standard values, see e.g. [16])
has disappeared (values in the range 55{65 km s 1 Mpc 1 are now obtained using this
data [12]).
The BOOMERANG maps have recently been searched for non-Gaussianity using
a pixel space analysis [17]. Non-Gaussianity was found at lower Galactic latitudes but
this is thought to be due to residual dust contamination. Limits at the level of a 2%
to 8% percent are found for the amount of an admixture of non-Gaussian component
(of a specic kind) that could be present in the higher latitude data.
4. MAXIMA
Shortly after the rst announcement of the BOOMERANG Antarctica results, results
were also announced from another balloon experiment, MAXIMA-1 [18]. This
experiment is not long-duration, but has a very sensitive array receiver. The result is
coverage over a eld somewhat smaller than that observed by BOOMERANG (about
10 by 10 degrees), resulting in less resolution in ` space, but with enough sensitivity to
make a highly signicant determination of the power spectrum, over a similar range.
The initial power spectrum from MAXIMA-1 is shown Fig. 2 in comparison with the
initial one from BOOMERANG. It tends to give a slightly higher rst peak, at slightly
higher `, again strongly favouring a at universe, but with a similar depression of the
second peak.
In a similar fashion to the reanalysis of the BOOMERANG data leading to
results at higher resolution, the MAXIMA-1 data have also been reanalysed, with
the results presented in Lee et al [19]. In particular this analysis extends the results
from 36  `  785 to ` < 1235. However, in contrast to the BOOMERANG case,
less data is used here than in the rst publication. This is because only the central
cross-linked area of the map has been used, and also some 240 GHz data was found
to be inconsistent for ` > 785. The result is that while the data provides useful high
resolution points extending out to where the third peak should be present, there is no
increase in statistical signicance compared to the rst results. Parameter estimation
using these new MAXIMA-1 results is discussed in Stompor et al [20], and an estimate
of the bispectrum using the MAXIMA-1 data is given in Santos et al [21], resulting in
weak constraints on non-Gaussianity.
These (reanalysed) MAXIMA-1 results are shown in comparison with the new
BOOMERANG points and those from DASI and CBI (see below) in the compilation
plot Fig. 6, which also includes the old MAXIMA and BOOMERANG points so that
a comparison can be made. Taken together, there is now overwhelming evidence for
the existence of the rst peak, but the status of peaks 2 and 3 is less clear. Future

CMB anisotropies: recent measurements and interpretation 10
Figure 6. Compilation of recent data (prepared by C. 
Odman).
data from interferometers that provide higher angular resolution than the balloon data
(e.g. CBI and VSA, see below) will be vital in sorting out the status of these and
subsequent peaks.
5. Further data from balloon ights
As already discussed, the next BOOMERANG ight (Dec 2002/Jan2003) will
concentrate on polarisation. This will use the same type of polarisation sensitive
bolometer (PSB) as will be own on the Plank mission (see below), thus providing
a valuable test of the Planck technology ahead of launch. Before this, there is a
ight scheduled for April 2002 in which a new version of MAXIMA, also adapted for
polarisation measurements, MAXIPOL, will be own. The MAXIMA and MAXIPOL
web pages [22, 23] contain further details.
Two further long duration balloon experiments have already own and rst results
from them are expected shortly. The rst is TOPHAT, [24]. TOPHAT is another long
duration balloon with multiple bolometers. Its rst (also last!) ight was in January
2001. The principal innovation with TOPHAT is to place the telescope on top of
the balloon. This provides good shielding from possible ground pick-up. TOPHAT
obtained 6 days of good data but then became becalmed over Antarctica Plateau,
and eventually had to be dropped in a region where recovery of the telescope itself

CMB anisotropies: recent measurements and interpretation 11
was impossible, although the scientic data was successfully salvaged. The main
analysis problem has been pointing reconstruction, in common with other LDB balloon
experiments, but in this case including a new torsional mode due to the surface of the
balloon itself.
Archeops [25] uses polar ballooning (similar to BOOMERANG), but in the
Northern Hemisphere, and relies on winds carrying it eastward from its launch point
in Kiruna in Sweden. Its main dierence from BOOMERANG is in its scanning
strategy { full sky circles instead of back and forth scans of about 60 ô . This clearly
gives less sensitivity per pixel, but on the other hand allows it to achieve improved
resolution in ` space on the power spectrum. Its (non-test) ights so far have been in
Jan 2000 and 2001 from Kiruna, with a further campaign planned for January 2002.
The three channels of the instrument are the same as the PLANCK HFI ones at 143,
217 and 353 GHz, with beams of 8.0, 5.5 and 5.0 arcmins (FWHM). It is therefore
a good preliminary test of Planck ideas and optics before launch. Eight bolometers
are available at each freq. A 24 hour ight was achieved in Jan 2000 and 7.5 hours
of good data are available from the Jan 2001 ight. Results from both Archeops and
TOPHAT are awaited with great interest.
6. The Degree Angular Scale Interferometer
DASI, the Degree Angular Scale Interferometer, is a 13 element interferometer working
at 30 GHz sited at South Pole. The results presented in April 2001 [26] used 97 days of
data covering 32 elds. Ground pickup, a big problem for this type of interferometer,
is subtracted by using rows of 8 elds 1 hour apart in Right Ascension observed over
the same azimuth range.
As mentioned for BOOMERANG, calibration is diôcult at the pole (e.g. Jupiter
is always near the horizon) but the DASI team report a nal calibration uncertainty
of approximately 3.5%. This should be compared with the gure of 10% for
BOOMERANG, which of course translates to 20% in C ` s. This highlights one
advantage of interferometers | the calibration is relatively simpler and more accurate
than with balloon experiments, so that we may expect in the future interferometers to
provide the best estimates of e.g. the rst peak height in the power spectrum. Also,
interferometers do not suer from lack of knowledge of the beam in same way as for
single dish balloon experiments | the power spectrum response is just a function of
the (u; v) visibility positions and these are known precisely! The major contaminant
is discrete radio sources { this is a problem for DASI due to the lack of measurements
of radio sources at higher frequencies in the Southern Hemisphere. They eectively
work by assuming source positions are known and then marginalizing over ux. This
method appears to work well at DASI resolutions, but actual measurements of the
sources' uxes at the working frequency are probably necessary when working at
higher resolution. The DASI results are shown in the composite plot Fig. 6, and
provide a good visual appearance of multiple peaks in the power spectrum, although,
as with BOOMERANG, the statistical signicance of peaks beyond the rst is only
about 2. Nevertheless, these are very impressive results. Cosmological parameter
estimation using the DASI data is presented in [27].

CMB anisotropies: recent measurements and interpretation 12
7. The Cosmic Background Interferometer
The CBI, or Cosmic Background Interferometer is a 13 element interferometer
currently sited in Chile. It is designed to be complementary to the `-space coverage
of DASI and also ts in well with the angular coverage of the VSA. In principle it can
reach an ` of 4000, although so far results from just two deep elds have been published,
resulting in two broad bins in ` space [28]. Shortly results from 100 mosaiced elds
will be available giving 10 bins, width 200 in `, out to ` = 2000, with the promise
therefore of very exciting results. More information and views of this telescope can be
found on the CBI webpage at [29], and on DASI at [30].
8. The Very Small Array
The Very Small Array (VSA) is 14-element interferometer currently operating in
Tenerife. It was built in a partnership between the Cavendish Laboratory Cambridge,
the Nuôeld Radio Astronomy Laboratory at Jodrell Bank (Manchester), and the IAC
Tenerife. It uses the same aperture synthesis techniques as DASI and CBI, although
with a dierent form of mounting in which the horn apertures can rotate independently
on their mountings, rather than all being co-mounted on the platform. This has
rendered it less susceptible to ground pickup than the other two instruments, although
pick up between the horns has to be ltered out on spacings corresponding to low `.
It observes at 26-36 GHz, from the Observatorio del Teide, Tenerife and is designed to
image and measure the power spectrum of CMB anisotropies between 100 < ` < 1800
This coverage is achieved by having two array congurations | a compact array,
corresponding to lower `, and an extended array for higher `. Temperature sensitivity
is maintained in the extended array by mounting larger horns in this case. A picture
of the instrument in the extended array conguration is shown in Fig. 7.
Observations in the compact array have now ended and the results from this are
expected very shortly, covering a range in ` out to about 900, and thus able to provide
important information on the existence or otherwise of the second and third peaks.
The VSA is now taking data in the extended array, and these promise to give good
sensitivity and accuracy in the important range 300 < ` < 1800. The problems faced
by both DASI and CBI with regard to source subtraction and dealt with by the VSA
by the use of a separate source subtraction interferometer co-sited with the VSA,
which is able to measure at 30 GHz sources detected at 15 GHz by the Ryle Telescope
in Cambridge. More information on the current status of the VSA and results can be
found on the VSA webpage, [31].
9. Results from parameter estimation
A great deal of work has been carried out already on extracting cosmological
parameters from current CMB data, and combining this with other cosmological data
sets. It is always necessary to stress that such work is relative to the particular
cosmological models which are adopted (the most popular being adiabatic Gaussian
perturbations from in ation), and is unable to treat systematics in the data beyond
those which are known and can be described in a convenient quantitative form (such
as the eects of calibration and beamwidth uncertainty). Furthermore, due to speed
limitations in being able to compute enough CMB power spectra quickly enough,
the model spaces considered have been somewhat restricted compared to the range

CMB anisotropies: recent measurements and interpretation 13
Figure 7. The Very Small Array (VSA) in its extended array conguration.
of parameters that might be expected. In particular, the treatments given in the
papers corresponding to individual experiments such as BOOMERANG, DASI and
MAXIMA ([12, 27, 20]) have tended to ignore a possible tensor component, and have
thus reached somewhat more restrictive conclusions that would otherwise be obtained.
That said, some remarkable results have already been found, strongly supporting
the in ationary paradigm for the origin of uctuations. Two papers which have
considered a wide range of data and models, including a tensor component, are
Wang et al [32] and Efstathiou et al [33], the latter of which combines CMB data
with matter power spectrum results from the 2dF galaxy survey. We use this latter
paper to show some illustrative results. In Fig. 8 likelihood plots for a subset of the
parameters are shown, at the top (labelled (a)) using CMB data alone and at the
bottom (labelled (c)) for the combination with 2dF. The parameters singled out here
are the departure of the universe from atness measured by the curvature parameter

k , the component in a cosmological
constant,
 , the Hubble constant h which is H 0
measured in units of 100 km s 1 Mpc 1 , the slope of the initial power law spectrum
for the scalar component, n s and the physical densities (i.e. after multiplication by
h 2 ), of the cold dark matter and baryonic components, ! c and ! b respectively. The
top set of plots illustrate clearly the degeneracies in parameter pairs which occur in
using just the CMB data alone. As already mentioned, these plots are somewhat less
restrictive than some which have appeared in individual experiments analysis, due to
the inclusion and marginalisation over an arbitrary tensor component of the spectrum.

CMB anisotropies: recent measurements and interpretation 14
Figure 8. Parameter estimation results from (top) CMB alone and (bottom)
combining the CMB data with the 2dF matter power spectrum information. 1, 2
and 3  condence regions are shown. (Taken from Efstathiou et al [33].)

CMB anisotropies: recent measurements and interpretation 15
Fortunately, the degeneracies in parameter estimation using the CMB are
generally orthogonal to those which occur when using other data sets, and shown
in the bottom panels of Fig. 8 are the same results when the matter power spectrum
likelihood function from the 2dF survey is included. The parameter estimation now
settles unambiguously on a
high
 , a reasonable h and a power law slope n s slightly
larger than but compatible with 1. It is interesting that the results
for
 here,
which after marginalisation over other parameters yield 0:65
<
 < 0:85 at 95%
condence, conrm the presence of a cosmological constant completely independently
of the supernovae data.
Increasing the dimensions of likelihood grids in order to estimate or marginalise
over more parameters is currently very expensive computationally. For example,
adding just one more parameter requiring 50 points to dene its functional form,
means an increase in the size of the grid and the time taken to compute it by a factor
50. This rapidly becomes prohibitive. An alternative is a Markov Chain Monte Carlo
technique, which does not work on a grid at all, but attempts to map out the likelihood
surface via well chosen random samples. This will almost certainly be very important
in future cosmological data tting, and has already been employed in a CMB context
by Knox et al [34]. These authors also draw attention a surprisingly tight constraint
on the age of the universe that can be drawn from current CMB data alone if one
assumes a at
universe(
tot = 1). This result is t 0 = 14:00:5 Gyr, and the relatively
small uncertainty is due to the tight correlation in at models between the age of the
universe and the angle subtended by the sound horizon at last scattering.
10. The future | observationally
10.1. Satellites
MAP (the Microwave Anisotropy Probe) was launched on June 30th, 2001, and is now
taking data at the second Lagrangian point (L2) of the Sun-Earth system. It will take
observations for a minimum of two years.
MAP has 5 frequency channels (22 to 90 GHz) and a best resolution of 12 arcmin.
It should thus be able to dene the CMB power spectrum accurately to the third
peak, limited mainly by cosmic variance over most of its ` range. The eventual
sensitivity after two years of coverage is expected to be 30 K per beam area. A
key benet from this mission will be all-sky multifrequency coverage, allowing for
denitive normalisation of the CMB power spectrum, as well as robust estimates of
contaminants in the sub 100 GHz range. MAP is discussed in more detail elsewhere
in this volume.
Planck is an ESA mission due for launch in 2007. Planck has ten frequency
channels spanning 30 to 800 GHz and a highest resolution of 5 arcmin. It should be
able to achieve an eventual sensitivity of about 5 K per beam area. It is expected
the CMB power spectrum will be recovered extremely accurately to approximately
the 8th peak.
The experimental parameters of the Planck mission are shown in Table 1.
The mission is designed to give high sensitivity to CMB structures, together with
suôcient frequency coverage to enable accurate separation of the non-CMB physical
components. These will typically be Galactic dust, synchrotron and free-free emission,
together with extragalatic radio and sub-mm/FIR sources. Also present will be
the eects of Sunyaev-Zeldovich distortion of the CMB as it passes through the

CMB anisotropies: recent measurements and interpretation 16
LFI (HEMT) HFI (Bolometers)
 GHz 30 44 70 100 100 143 217 353 545 857
No. of detectors 4 6 12 34 4 12 12 6 8 6
 FWHM 33 0 23 0 14 0 10 0 10:7 0 8 0 5:5 0 5 0 5 0 5 0

T=T  10 6 1.6 2.4 3.6 4.3 1.7 2.0 4.3 14.4 147 6670
Polarization yes yes yes yes no yes yes no yes no
Table 1. Experimental parameters of the Planck satellite. HFI refers to the high
freqquncy component of the instrument package, and LFI is the low frequency
instrument. The
T=T sensitivity is per beam area in one year (thermodynamic
temperature).
hot intracluster gas of clusters of galaxies (see below). Schemes for achieving this
separation (e.g. using the Maximum Entropy Method, as in [35]), typically show that
the required accuracy can probably be achieved over the majority of the sky, and as
an extremely signicant byproduct, catalogues of extragalactic sources and clusters of
galaxies will also be produced.
The satellite will spin at about 1 rpm with the detector beams tracing out circles
quite close to great circles, in a pattern which slowly precesses to cover the full sky once
every 6 months. Thus two full sky coverages are expected over the 1 year course of the
mission. The main systematic eects expected for the Planck data, aside from those
due to Galactic and extragalactic sources, will arise from `scanning eects' around the
scan circles, due to variations in instrument noise and d.c. levels. Schemes for removal
of these scan eects and the reconstruction of the sky using the total set of scan circles
are now well advanced, and there is good reason to hope that the Planck results will
reach close to the levels of accuracy indicated by the noise sensitivity projections in
Table 1.
A crucial aspect of the Planck mission, will be the extent to which it will be able
to recover polarization information on the CMB. Polarization sensitive bolometers will
be used in 3 of the High Frequency Instrument (HFI) channels, and all 4 of the Low
Frequency Instrument (LFI) HEMT channels can discriminate polarized radiation.
The projected polarization sensitivity should be suôcient to make measurement of
the B component of the polarization power spectrum measurable, with a consequent
ability to unambiguously separate a tensor mode from scalar mode contributions [36].
However, the extent to which foreground contamination compromises the achievable
accuracy has yet to be determined. For further information on the Planck project, see
e.g. the Planck Science Team web page [37].
10.2. Future ground-based experiments
The emphasis with ground-based experiments is now likely to shift to higher ` and the
measurement of polarization, for both of which interferometers are likely to be very
important. At higher `, while the primordial CMB power spectrum itself is expected
to decline, several new secondary eects become important. The foremost amongst
these is the Sunyaev-Zeldovich (SZ) eect, a distortion of the CMB spectrum due
to passing through the hot intracluster gas of an intervening galaxy cluster. This
distortion manifests itself as a spatially modulated intensity change in the CMB |
negative in the Rayleigh-jeans region of the spectrum and positive in the Wien region.
An extremely important aspect of the SZ eect is that, for given cluster parameters,

CMB anisotropies: recent measurements and interpretation 17
it is independent of redshift, and therefore can provide a probe of mass concentrations
to high redshift.
Some blank sky searches for the SZ eect are now starting using bolometer arrays
(e.g. ACBAR and BOLOCAM) and construction is just starting for interferometers
also aimed at reaching higher angular resolution. These include the Arcminute
MicroKelvin Imager (AMI) in the U.K., the Sunyaev-Zeldovich Array (SZA) in the
U.S.A. and the AMIBA telescope in Taiwan. These new instruments will give the
rst unbiassed surveys of mass concentrations > 10 14 M out to large redshift | an
extremely important task. Since the abundance of clusters at high redshift is a very
sensitive function of cosmology, these surveys will provide estimates
of
m
and

quite independent of the results from the primordial CMB and supernovae searches
[38].
As an example, AMI will consist of ten 3.7 m dishes to be sited in Cambridge
U.K. in conjunction with an enhanced Ryle Telescope (eight 12.5 m dishes). It will
observe at 15 GHz, where low Galactic contamination is expected on these scales, and
will concentrate on blank eld Sunyaev-Zeldovich surveys. Work is already underway
on this project, with rst observations expected during 2003.
11. Conclusions
The following is a summary of the main conclusions which we have been able to draw
from the CMB so far:
 The rst (main) peak in the CMB power spectrum has now been measured and
there is evidence for further peaks.
 The universe is close to at.
 The density of matter in the universe (CDM and baryons) is about 0.2 to 0.3,
with baryons accounting for 15% to 20% of this.
 The remainder is dark energy { either the cosmological constant or eects of a
scalar eld (the latter being known as quintessence).
 There is strong evidence for this last conclusion independently from the
combination of either Supernovae and CMB or from combining Large Scale
Structure information and the CMB.
 The Hubble constant is about 70 km s 1 Mpc 1 and the age of the universe is
13{15 Gyr.
 The statistics of the CMB temperature uctuation eld are Gaussian to a level
of a few percent.
These are obviously all very impressive achievements for recent cosmological
investigation, and overall conrm that in ation is doing very well as a predictive
theory! On the downside, the nature of the dark matter is still unknown (it is possibly
not even particles but represents something wrong with our current understanding of
gravity) and of course explaining the presence of a small but non-zero cosmological
constant remains a big challenge.
Acknowledgments
I would like to thank the organisers very much for the opportunity to attend the
excellent conference in Capetown.

CMB anisotropies: recent measurements and interpretation 18
References
[1] R.K. Sachs and A.M. Wolfe. Astrophys.J., 147:73, 1967.
[2] C.M. Gutierrez, R. Rebolo, R.A. Watson, R.D. Davies, A.W. Jones, and A.N. Lasenby. The
Tenerife Cosmic Microwave Background maps: observations and rst analysis. Astrophys.J.,
529:47, 2000.
[3] C.B. Nettereld, M.J. Devlin, N. Jarosik, L. Page, and E.J. Wollack. A measurement of the
angular power spectrum of the anisotropy in the cosmic microwave background. Astrophys.
J., 474:47, 1997.
[4] A.N. Lasenby. Observations of the Cosmic Microwave Background. In R.G. Crittenden and
N.G. Turok, editors, Structure formation in the Universe: NATO ASI, Cambridge, U.K.,
1999, volume 565, page 215. Kluwer: NATO Science Series, 2001.
[5] P.D. Mauskopf, P.A.R. Ade, P. de Bernardis, J.J. Bock, J. Borrill, A. Boscaleri, B.P. Crill,
G. DeGasperis, G. De Troia, P. Farese, P.G. Ferreira, K. Ganga, M. Giacometti, S. Hanany,
V.V. Hristov, A. Iacoangeli, A.H. Jae, A.E. Lange, A.T. Lee, S. Masi, A. Melchiorri,
F. Melchiorri, L. Miglio, T. Montroy, C.B. Nettereld, E. Pascale, F. Piacentini, P.L.
Richards, G. Romeo, J.E. Ruhl, E. Scannapieco, F.Scaramuzzi, R. Stompor, and N. Vittorio.
Measurement of a peak in the cosmic microwave background power spectrum from the North
American test ight of BOOMERANG. Submitted to Astrophys.J., astro-ph/9911444, 1999.
[6] A. Melchiorri, P.A.R. Ade, P. de Bernardis, J.J. Bock, J. Borrill, A. Boscaleri, B.P. Crill, G. De
Troia, P. Farese, P.G. Ferreira, K. Ganga, G. de Gasperis, M. Giacometti, V.V. Hristov,
A.H. Jae, A.E. Lange, S. Masi, P.D. Mauskopf, L. Miglio, C.B. Nettereld, E. Pascale,
F. Piacentini, G. Romeo, J.E. Ruhl, and N. Vittorio. A measurement of Omega from the
North American test ight of BOOMERANG. Submitted to Astrophys.J., astro-ph/9911445,
1999.
[7] P. de Bernardis, P.A.R. Ade, J.J. Bock, J.R. Bond, J. Borrill, A. Boscaleri, K. Coble, B.P. Crill,
G. De Gasperis, P.C. Farese, P.G. Ferreira, K. Ganga, M. Giacometti, E. Hivon, V.V. Hristov,
A. Iacoangeli, A.H. Jae, A.E. Lange, L. Martinis, S. Masi, P.V. Mason, P.D. Mauskopf,
A. Melchiorri, L. Miglio, T. Montroy, C.B. Nettereld, E. Pascale, F. Piacentini, D. Pogosyan,
S. Prunet, S. Rao, G. Romeo, J.E. Ruhl, F. Scaramuzzi, D. Sforna, and N. Vittorio. A at
universe from high-resolution maps of the cosmic microwave background radiation. Nature,
404:955, 2000.
[8] A.E. Lange, P.A.R. Ade, J.J. Bock, J.R. Bond, J. Borrill, A. Boscaleri, K. Coble, B.P. Crill,
P. de Benardis, P. Farese, P. Ferreira, K. Ganga, M. Giacometti, E. Hivon, V. V. Hristov,
A. Iacoangeli, A.H. Jae, L. Martinis, S. Masi, P.D. Mauskopf, A. Melchiorri, T. Montroy,
C.B. Nettereld, E. Pascale, F. Piacentini, D. Pogosyan, S. Prunet, S. Rao, G. Romeo, J.E.
Ruhl, F. Scaramuzzi, and D. Sforna. First estimations of cosmological parameters from
BOOMERANG. Submitted, astro-ph/0005004, 2000.
[9] M. Tegmark and M. Zaldarriaga. New CMB constraints on the cosmic matter budget: trouble
for nucleosynthesis? Submitted to Astrophys.J., astro-ph/0004393, 2000.
[10] R.A. Battye, R. Crittenden, and J. Weller. Cosmic concordance and the ne structure constant.
Phys.Rev.D., 6304(4):3505, 2001.
[11] P.R. Avelino, S. Esposito, G. Mangano, C.J.A.P. Martins, A. Melchiorri, G. Miele, O. Pisanti,
G. Rocha, and R.T.R. Viana. Early-universe constraints on a time-varying ne structure
constant. Phys.Rev.D., 6410:3505, 2001.
[12] C.B. Nettereld, P.A.R. Ade, J.J. Bock, J.R. Bond, J. Borrill, A. Boscaleri, K. Coble, C.R.
Contaldi, B.P. Crill, P. de Bernardis, P. Farese, K. Ganga, M. Giacometti, E. Hivon,
V.V. Hristov, A. Iacoangeli, A.H. Jae, W.C. Jones, A.E. Lange, L. Martinis, S. Masi,
P. Mason, P.D. Mauskopf, A. Melchiorri, T. Montroy, E. Pascale, F. Piacentini, D. Pogosyan,
F. Pongetti, S. Prunet, G. Romeo, J.E. Ruhl, and F. Scaramuzzi. A measurement by
BOOMERANG of multiple peaks in the angular power spectrum of the cosmic microwave
background. Submitted to Astrophys.J., astro-ph/0104460, 2001.
[13] S.L.Bridle, R.Crittenden, A.Melchiorri, M.P.Hobson, R.Kneissl, and A.N. Lasenby. An-
alytic marginalization over CMB calibration and beam uncertainty. Submitted to
Mon.Not.R.astr.Soc., astro-ph/0112114, 2001.
[14] P. de Bernardis, P. A. R. Ade, J. J. Bock, J. R. Bond, J. Borrill, A. Boscaleri, K. Coble, C. R.
Contaldi, B. P. Crill, G. De Troia, P. Farese, K. Ganga, M. Giacometti, E. Hivon, V. V.
Hristov, A. Iacoangeli, A. H. Jae, W. C. Jones, A. E. Lange, L. Martinis, S. Masi, P. Mason,
P. D. Mauskopf, A. Melchiorri, T. Montroy, C. B. Nettereld, E. Pascale, F. Piacentini,
D. Pogosyan, G. Polenta, F. Pongetti, S. Prunet, G. Romeo, J. E. Ruhl, and F. Scaramuzzi.
Multiple Peaks in the Angular Power Spectrum of the Cosmic Microwave Background:

CMB anisotropies: recent measurements and interpretation 19
Signicance and Consequences for Cosmology. Astrophys.J., 564:559, 2002.
[15] S. Burles, K. M. Nollett, and M. S. Turner. Big Bang Nucleosynthesis Predictions for Precision
Cosmology. Astrophys.J.Letts., 552:L1, 2001.
[16] O. Lahav, S.L. Bridle, M.P. Hobson, A.N. Lasenby, and L. Sodre. Bayesian hyper-
parameters approach to joint estimation: The Hubble constant from CMB measurements.
Mon.Not.R.astr.Soc., 315:L45, 2000.
[17] G. Polenta, P.A.R. Ade, J.J. Bock, J.R. Bond, J. Borrill, A. Boscaleri, C.R. Contaldi, B.P.
Crill, P. de Bernardis, G. De Gasperis, G. De Troia, K. Ganga, M. Giacometti, E. Hivon,
V.V. Hristov, A.H. Jae, A.E. Lange, S. Masi, P.D. Mauskopf, A. Melchiorri, T. Montroy,
P. Natoli, C.B. Nettereld, E. Pascale, F. Piacentini, D. Pogosyan, S. Prunet, G. Romeo,
J.E. Ruhl, N. Vittorio, and A.Zeppilli. Search for non-gaussian signals in the BOOMERanG
maps: pixel-space analysis. Submitted to Astrophys.J.Letts., astro-ph/0201133, 2002.
[18] S. Hanany, P. Ade, A. Balbi, J. Bock, J. Borrill, A. Boscaleri, P. de Bernardis, P. G. Ferreira,
V. V. Hristov, A. H. Jae, A. E. Lange, A. T. Lee, P. D. Mauskopf, C. B. Nettereld, S. Oh,
E. Pascale, B. Rabii, P. L. Richards, G. F. Smoot, R. Stompor, C. D. Winant, and J. H. P. Wu.
MAXIMA-1: A Measurement of the Cosmic Microwave Background Anisotropy on Angular
Scales of 10 0 {5 ô . Astrophys.J.Letts., 545:L5, 2000.
[19] A. T. Lee, P. Ade, A. Balbi, J. Bock, J. Borrill, A. Boscaleri, P. de Bernardis, P. G. Ferreira,
S. Hanany, V. V. Hristov, A. H. Jae, P. D. Mauskopf, C. B. Nettereld, E. Pascale, B. Rabii,
P. L. Richards, G. F. Smoot, R. Stompor, C. D. Winant, and J. H. P. Wu. A High Spatial
Resolution Analysis of the MAXIMA-1 Cosmic Microwave Background Anisotropy Data.
Astrophys.J., 561:L1, 2001.
[20] R. Stompor, M. Abroe, P. Ade, A. Balbi, D. Barbosa, J. Bock, J. Borrill, A. Boscaleri, P. de
Bernardis, P. G. Ferreira, S. Hanany, V. Hristov, A. H. Jae, A. T. Lee, E. Pascale, B. Rabii,
P. L. Richards, G. F. Smoot, C. D. Winant, and J. H. P. Wu. Cosmological Implications of
the MAXIMA-1 High-Resolution Cosmic Microwave Background Anisotropy Measurement.
Astrophys.J.Letts., 561:L7, 2001.
[21] M.G. Santos, A. Balbi, J. Borrill, P.G. Ferreira, S. Hanany, A.H. Jae, A.T. Lee, J. Magueijo,
B. Rabii, P.L. Richards, G.F. Smoot, R. Stompor, C.D. Winant, and J.H.P. Wu. An
estimate of the Cosmological Bispectrum from the MAXIMA-1 CMB map. Submitted to
Phys.Rev.Letts., astro-ph/0107588, 2001.
[22] http://cosmology.berkeley.edu/group/cmb/index.html.
[23] http://www.physics.umn.edu/cosmology/maxipol/.
[24] http://topweb.gsfc.nasa.gov/.
[25] http://archeops01.free.fr/main_archeops/index_english.html.
[26] N.W. Halverson, E.M. Leitch, C. Pryke, J. Kovac, J.E. Carlstrom, W.L. Holzapfel, M. Dragovan,
J. K. Cartwright, B.S. Mason, S. Padin, T.J. Pearson, M.C. Shepherd, and A.C.S. Readhead.
DASI rst results: A measurement of the cosmic microwave background angular power
spectrum. Submitted to Astrophys.J., astro-ph/0104489, 2001.
[27] C. Pryke, N.W. Halverson, E.M. Leitch, J. Kovac, J.E. Carlstrom, W.L. Holzapfel, and
M. Dragovan. Cosmological Parameter Extraction from the First Season of Observations
with the Degree Angular Scale Interferometer. Astrophys.J., 568:46, 2002.
[28] S. Padin, J. K. Cartwright, B. S. Mason, T. J. Pearson, A. C. S. Readhead, M. C. Shepherd,
J. Sievers, P. S. Udomprasert, W. L. Holzapfel, S. T. Myers, J. E. Carlstrom, E. M. Leitch,
M. Joy, L. Bronfman, and J. May. First Intrinsic Anisotropy Observations with the Cosmic
Background Imager. Astrophys.J., 549:L1, 2001.
[29] http://www.astro.caltech.edu/~tjp/CBI/.
[30] http://astro.uchicago.edu/dasi/.
[31] http://www.mrao.cam.ac.uk/telescopes/vsa/.
[32] X. Wang, M. Tegmark, and M. Zaldarriaga. Is cosmology consistent? To be published in
Phys.Rev.D., astro-ph/0105091, 2001.
[33] G. Efstathiou, S. Moody, J.A. Peacock, W.J. Percival, C. Baugh, J. Bland-Hawthorn, T. Bridges,
R. Cannon, S. Cole, M. Colless, C. Collins, W. Couch, G. Dalton, R. de Propris, S.P. Driver,
R.S. Ellis, C.S. Frenk, K. Glazebrook, C. Jackson, O. Lahav, I. Lewis, S. Lumsden, S. Maddox,
P. Norberg, B.A. Peterson, W. Sutherland, and K. Taylor. Evidence for a non-zero  and a
low matter density from a combined analysis of the 2dF Galaxy Redshift Survey and Cosmic
Microwave Background anisotropies. Mon.Not.R.astr.Soc., 330:L29, 2002.
[34] L. Knox, N. Christensen, and C. Skordis. The Age of the Universe and the Cosmological
Constant Determined from Cosmic Microwave Background Anisotropy Measurements.
Astrophys.J.Lett., 563:L95, 2001.
[35] M.P. Hobson, A.W. Jones, A.N. Lasenby, and F.R. Bouchet. Component separation methods

CMB anisotropies: recent measurements and interpretation 20
for satellite observations of the cosmic microwave background. Mon.Not.R.astr.Soc., 300:1,
1998.
[36] M. Zaldarriaga and U. Seljak. An all-sky analysis of polarization in the microwave background.
Phys.Rev.D., 55:1830, 1997.
[37] http://astro.estec.esa.nl/Planck/.
[38] R. Kneissl, M. E. Jones, R. Saunders, V. R. Eke, A. N. Lasenby, K. Grainge, and G. Cotter.
Surveying the sky with the Arcminute MicroKelvin Imager: expected constraints on galaxy
cluster evolution and cosmology. Mon.Not.R.astr.Soc., 328:783, 2001.