a

r
X
i
v
:
h
e
p
-
p
h
/
0
5
0
3
2
4
6
v
1


2
4

M
a
r

2
0
0
5
hep-ph/0503246 IFUPTH/200506
Implications of neutrino data circa 2005
Alessandro Strumia
Dipartimento di Fisica dellUniversit` a di Pisa and INFN
Francesco Vissani
INFN, Laboratori Nazionali del Gran Sasso, Theory Group, I-67010 Assergi (AQ), Italy
Abstract
Adopting the 3 neutrino framework, we present an updated determination of the
oscillation parameters. We perform a global analysis and develop simple arguments
that give essentially the same result. We also discuss determinations of solar neutrino
uxes, capabilities of future experiments, tests of CPT, implications for neutrino-less
double- decay, decay, cosmology.
Oscillation parameter central value 99% CL range
solar mass splitting m
2
12
= (8.0 0.3) 10
5
eV
2
(7.2 8.9) 10
5
eV
2
atmospheric mass splitting [m
2
23
[ = (2.5 0.3) 10
3
eV
2
(1.7 3.3) 10
3
eV
2
solar mixing angle tan
2

12
= 0.45 0.05 30

<
12
< 38

atmospheric mixing angle sin

2
2
23
= 1.02 0.04 36

<
23
< 54

CHOOZ mixing angle sin

2
2
13
= 0 0.05
13
< 10

non-oscillation probed experimental 99% CL range 99% CL range

parameter by limit at 99% CL normal hierarchy inverted hierarchy
ee-entry of m 02 m
ee
< 0.38 heV (1.1 4.5) meV (12 57) meV
(m

m)
1/2
ee
-decay m
e
< 2.0 eV (4.6 10) meV (42 57) meV
m
1
+m
2
+m
3
cosmology m
cosmo
< 0.94 eV (51 66) meV (83 114) meV
Table 1: Summary of present information on neutrino masses and mixings from oscillation data (upper
rows) and inferences and limits on non-oscillation probes (lower rows) A 99% C.L. range is a 2.58
range. Some 02 data are controversial, and h 1 parameterizes uncertain nuclear matrix elements.
10
-5
10
-4
10
-3
10
-2
m
2
in eV
2
0
2
4
6
8
10
12
14

2
Neutrino masses
m
12
2
| m
23
2
|
90 %CL
99 %CL
0 10 20 30 40 50 60
in degree
0
2
4
6
8
10
12
14

2
Neutrino mixings

12

23

13
90 %CL
99 %CL
Figure 1: Summary of present information on neutrino masses and mixings from oscillations.
1
The most plausible extension of the Standard Model that allows to interpret a wealth of neutrino
data [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11] consists in adding a Majorana mass term for neutrinos,
L = L
SM
+
1
2
( m + h.c.) m = V

diag(m
1
, m
2
e
2i
, m
3
e
2i
) V

. (1)
For normal mass hierarchy we order the neutrino masses m
1,2,3
0 as m
1
< m
2
< m
3
, whereas for
inverted mass hierarchy we choose m
3
< m
1
< m
2
. The neutrino mixing matrix is
V = R
23
(
23
) diag(1, e
i
, 1) R
13
(
13
) R
12
(
12
). (2)
A similar result holds in the case of Dirac masses, with the dierence that the number of physical
parameters decreases from 9 to 7: the Majorana phases and can be reabsorbed by eld redenitions.
Data on neutrino oscillations x
12
,
23
, m
2
12
and [m
2
23
[ where m
2
ij
m
2
j
m
2
i
. As discussed
later (sections 1 and 2) our present knowledge of oscillation parameters is approximatively summarized
in the upper rows of table 1. The uncertainties are almost Gaussian in the chosen variables; g. 1
shows the full
2
functions. Correlations among parameters are ignored because negligible, with the
exception that the upper bound on
13
depends on [m
2
23
[. While
12
is rather precisely measured,
the other 2 mixing angles have large uncertainties. Planned long-baseline oscillation experiments can
strongly improve on m
2
23
and (if
13
>

) measure
13
, the phase , and determine which type of
mass hierarchy is realized in nature.
Oscillation experiments, however, are insensitive to the absolute neutrino mass scale (say, the mass
of the lightest neutrino) and to the 2 Majorana phases and . Other types of experiments can study
some of these quantities and the nature of neutrino masses. They are: -decay experiments, that in
good approximation probe m
2
e
(m m

)
ee
=

i
[V
2
ei
[m
2
i
; neutrino-less double-beta decay (02)
experiments, that probe the absolute value of the Majorana mass m
ee

i
V
2
ei
m
i
; cosmological
observations (Large Scale Structures and anisotropies in the Cosmic Microwave Background), that in
good approximation probe m
cosmo

h
2
93.5eV =

i
m
i
. Only neutrinoless double beta decay
(02) probes the Majorana nature of the mass. The values [m
ee
[, m
e
, m
cosmo
are unknown, but
can be partially inferred from oscillation data. Table 1 shows our results, discussed in section 3, in
the limit where the mass of the lightest neutrino is negligible. In the opposite limit neutrinos are
quasi-degenerate and [m
ee
[, m
e
, m
cosmo
can be arbitrarily large.
From the point of view of 3 massive neutrinos, it is natural to divide in three parts a discussion of
the present situation and of the perspectives of improvements, namely:
1. Oscillations with solar frequency, that tell m
2
12
and
12
and give a sub-dominant constraint on

13
. In section 1 we discuss solar and reactor neutrino experiments, showing that the program
of measurement of parameters is well under way (if not accomplished), and discussing other
interesting related goals.
2. Oscillations with atmospheric frequency, that tell m
2
23
,
23
and
13
, are discussed in section 2.
3. Non-oscillation experiments, that can tell the absolute neutrino mass. In section 3 we discuss
the present status and assess the implications of the existing information on neutrino oscillations
for these experiments, particularly for 02. We conclude by commenting on the recent claim
of evidence for this transition [12, 13].
We assume the 3 neutrino framework because it is plausible, well dened, restrictive and compatible
with data. However it is just an assumption, and before proceeding we recall some alternatives. The
most plausible one is the presence of extra light fermions (sterile neutrinos) or bosons, which can
manifest in many ways. Going to rather exotic scenarios, Lorentz or CPT invariance (here studied
2
in g. 3) might be violated in neutrinos, that might have anomalous interactions (gauge couplings,
or magnetic moments, or else), might not obey the Pauli principle, etc, etc, etc. Present solar and
atmospheric data cannot be explained by these alternatives, which however might be present as sub-
leading eects on top of oscillations among active neutrinos, such that our determinations of active
oscillation parameters would need model-dependent modications. The present data do not give
any clear indication for extra eects, but contain some anomalous hints. Most notably, the LSND
anomaly [14] is not compatible with the 3 neutrino context we assume.
1 Oscillations with solar frequency
A few years ago the solar anomaly rested on global ts that combined solar model predictions with a
few measurements of solar neutrino rates. In recent times, the situation changed. As prospected in [15]
(written a few years ago, while sub-MeV solar experiments were discussed as a tool for discriminating
LMA from LOW, SMA, QVO,. . . ), SNO, KamLAND, Borexino had in any case the capability to
identify the true solution of the solar anomaly and make precision measurements of the oscillation
parameters. This is where we are now. From the point of view of the determination of the oscillation
parameters, solar neutrino experiments are in a more advanced stage than atmospheric experiments,
as clear from g. 1. KamLAND and SNO play the key r ole in the determination for m
2
12
and
12
respectively, and almost achieved the 2.5% and 10% accuracy prospected in [15]. In section 1.1 we
present a global solar t, and in section 1.2 we show that it is dominated by very simple and robust
inputs. In section 1.3 we extract from data the survival probability of low-energy neutrinos, and use
it to study how solar data mildly restrict
13
. In section 1.4 we reassess the interest in proceeding
with low-energy solar neutrino experiments. In section 1.5 we discuss how well present solar data
determine solar neutrino uxes and discuss the impact of Borexino. In all cases we extract from
simple arguments the main general results, that we compare with exact results of global ts performed
in specic cases.
1.1 Updated t of solar and reactor neutrino data
We begin by presenting a global t of solar and reactor neutrino data assuming oscillations among
active neutrinos with negligible
13
. We include
The nal SNO CC, NC, ES rates measured during day and during night without [6] and with [7]
salt, that gives enhanced NC sensitivity.
The Super-Kamiokande ES spectra [5].
The Gallium rate, R
Ga
= (68.13.7) SNU, obtained averaging the most recent SAGE data with
the nal Gallex and GNO data [2, 3, 4].
The Chlorine rate [1], R
Cl
= (2.56 0.23) SNU.
The KamLAND reactor anti-neutrino data with prompt energy higher than 2.6 MeV [8].
Solar model predictions and uncertainties [16] are revised including the recent measurement of the
14
N(p, )
15
O nuclear cross section [17], which reduces the predicted CNO uxes by roughly 50%.
The result of our oscillation t is shown in g. 2a, where the best t point is marked with a dot.
The 1 (i.e. 68% C.L.) and the 99% C.L. (i.e. 2.58) ranges for the single parameters (1 dof) are
summarized in table 1. The total evidence for an eect is now about 12 in solar data and about
3
0 0.2 0.4 0.6 0.8 1
tan
2

0
5
10
15
20

m
2

i
n
1
0
-
5
e
V
2
Reactor
_
Solar
90, 99, 99.73% CL
(2 dof)
0.2 0.25 0.3 0.35 0.4 0.45
P(
e

e
) at large energy
0.4
0.45
0.5
0.55
0.6
0.65
0.7
0.75
P
(

e
)
a
t
s
m
a
l
l
e
n
e
r
g
y

1
3
=
0

1
3
=
0
.
1

1
3
=
0
.
2

1
2
=
3
2
o

1
2
=
3
7
o
90, 99, 99.73% CL (2 dof)
Figure 2: Best-t regions at 90, 99 and 99.73% CL. Fig. 2a assumes CPT invariance and combines
solar data (dashed red contours) with reactor data (dotted blue contours). In g. 2a we show how
data determine the high- and low-energy limits of P(
e

e
), as precisely described in the text,
6 in KamLAND data. Fig. 2a also shows separately the t of solar data (dashed red contours) and
the t of KamLAND data (dotted blue contours).
We note in passing that, according to the Pearson
2
goodness-of-t (gof) employed by many old
global analyses, today the LOW solution (with m
2
12
10
7
eV
2
) still has a good gof. This does not
mean that LOW is compatible with data, but happens because the Pearson test is not a statistically
powerful gof test when dof 1 [18]: global ts of solar data input roughly 100 experimental inputs.
As we now discuss, the output depend almost only on a few pieces of data.
1.2 The meaning of global ts
Our results are based on a careful global t. We now point out how a simple approximate analysis is
sucient to get results practically equivalent to those of global ts.
The solar mass splitting m
2
12
is directly determined by the position of the oscillation dips at
KamLAND, with negligible contribution from solar experiments. (More precisely, this will be rigor-
ously true in the future. For the moment solar data are needed to eliminate spurious solutions mildly
disfavored by KamLAND data, as illustrated in g. 2a. Solar data have practically no impact around
the global minimum of the
2
, and consequently on the determination of m
2
12
as reported in table 1).
4
10
-3
10
-2 3 3 3
m
_
23
2
in eV
2
10
-3
10
-2
3
3
3

m
2
3
2

i
n
e
V
2
atmospheric oscillations
68,90,99% C.L. (2 dof)
C
P
T
c
o
n
s
e
r
v
i
n
g
10
-5
10
-4
10
-3
m
_
12
2
in eV
2
10
-5
10
-4
10
-3

m
1
2
2

i
n
e
V
2
solar oscillations
68,90,99% C.L. (2 dof)
C
P
T
c
o
n
s
e
r
v
i
n
g
Figure 3: Test of CPT-violating neutrino masses. We show the separate t for m
2
in neutrinos and
m
2
in anti-neutrinos, marginalized with respect to the mixing angles
12
and
23
. The atmospheric
t includes data from SK, K2K, Macro. The solar t includes data from SNO, SK, Gallex, Sage,
Homestake, KamLAND. These plots update the original results in g.s 5 and 6 of [19]
The solar mixing angle is directly determined by SNO measurements of NC and CC solar Boron
rates. Assuming avour conversions among active neutrinos, SNO implies
1
P(
e

e
)) (
e
)/(
e,,
) = 0.357 0.030.
This should be compared with the theoretical prediction for P(
e

e
)), given by a simple ex-
pressions that does not depend on the solar density prole because LMA oscillations are almost
completely adiabatic. At E

MeV matter eects dominate such that

e
produced around the
center of the sun coincide with the
2
eigenstate in matter and exit as the
2
eigenstate in vacuum,
so that P(
e

e
) sin
2

12
. In the energy range explored by SNO, matter eects at the production
region are not fully dominant, such that the above approximation gets slightly corrected to
2
P(
e

e
)) 1.15 sin
2

12
so that tan
2

12
= 0.45 0.05 (3)
which agrees with the results of the global analysis in table 1, both in the central value and in its
uncertainty.
Notice that the only solar model input that enters our approximate determination of solar oscilla-
tion parameters is the solar density around the center of the sun, that controls the 15% correction to
1
SNO measured (e,, ) = (4.90.3) 10
6
/cm
2
sec and (e) = (1.740.08) 10
6
/cm
2
sec. Each measurement was rst
performed using heavy water (CC and NC events mainly distinguished by their energy spectrum) and later with salt heavy
water (distinction relies on event isotropy). The measurements performed with these dierent experimental techniques
agree. Taking into account the SK measurement of the ES rate, (e) + 0.155(, ) = (2.35 0.06) 10
6
/cm
2
sec, the
SNO measurement of the ES rate, (e) + 0.155(, ) = (2.36 0.19) 10
6
/cm
2
sec and the solar model prediction
(e,, ) = (5.05 0.86) 10
6
/cm
2
sec would only marginally improve the measurement of P(e e) to P(e
e) = 0.360 0.028.
2
The factor giving the correction to P(e e) = sin
2
12 ranges between 1.1 and 1.2 within the present best-t
region at 90% CL.
5
P(
e

e
)) in eq. (3). This correction factor is comparable to the 1 uncertainty in P(
e

e
)):
indeed the associated increase of P(
e

e
) at smaller E

has not been observed in SNO and SK

spectra. For the reasons explained above the solar model independent t of [20] gives now a result
almost identical to the standard t, so that we do not show the update of this result.
Global ts remain still useful for testing if the pieces of data not included in our simplied analysis
(that have a minor impact in the standard t) contain statistically signicant indications for new
physics beyond LMA oscillations. At the moment the answer is no. E.g., g. 3b updates the CPT-
violating solar t of [19]: the best-t region includes the CPT-conserving limit (diagonal dotted line).
1.3 Eects of
13
and low-energy neutrinos
At low E

matter eects are negligible and the survival probability is given by averaged vacuum
oscillations. LMA oscillations with
13
= 0 predict the low-energy limit of P(
e

e
) in terms of its
high-energy limit as
3
P(
e

e
, small E

) = 1 2P(
e

e
, large E

) + 2P(
e

e
, large E

)
2
. (4)
A non zero
13
as well as new physics beyond neutrino masses allow to avoid this prediction. It is
therefore interesting to extract these two ideal observables from data. As discussed in the previous
section P(
e

e
, large E

) is presently dominantly determined by SNO. The low energy limit of

P(
e

e
) is presently dominantly determined by Gallium data and can be extracted by a simple
approximate argument [15, 4]. Subtracting from the total Gallium rate
(68.1 3.7) SNU = R
Ga
= R
Ga
pp,pep
+R
Ga
CNO
+R
Ga
7
Be
+R
Ga
8
B
(5)
its
8
B contribution (as directly measured by SNO via CC, R
Ga
8
B
= 4.3 1 SNU) and regarding all
remaining uxes as low energy ones, suppressed by P(
e

e
, small E

), determines it to be 0.57
0.03. Alternatively, by subtracting also the intermediate-energy CNO and Beryllium uxes, one gets
P(
e

e
, small E

) = 0.580.05. We here included only the error on the Gallium rate, which is the
dominant error. This rough analysis shows that the result only mildly depends on how one deals with
intermediate energy neutrinos, and on model-dependent details of the intermediate region, thereby
suggesting the following general result:
P(
e

e
, large E

) = 0.31 0.03, P(
e

e
, small E

) = 0.58 0.04. (6)

As illustrated in g. 4 at E

= 1 MeV (10 MeV) the dierence between the exact LMA prole of
P(
e

e
) and its low energy (high energy) limit is smaller than the 1 uncertainties in eq. (6). High
energy neutrinos have been dominantly measured by SNO at energies around 10 MeV.
In order to establish the validity of eq. (6) we compare it with the exact result of a global analysis
of solar data: to perform it we must abandon generality and focus on a specic mechanism that allows
to avoid the LMA prediction of eq. (4). We consider the case of a non vanishing
13
(see also [21]),
which gives
P(
e

e
, large E

) = sin
4

13
+ cos
4

13
sin
2

12
, (7a)
P(
e

e
, small E

) =

i
[V
ei
[
4
= sin
4

13
+ cos
4

13

1
1
2
sin
2
2
12

. (7b)
3
The transition between the two regimes proceeds at E few MeV, and the high-energy regime is approached
for E
>

20 MeV. At higher energies solar matter eects become comparable to the atmospheric mass splitting, giving
corrections to solar neutrino rates proportional to
2
13
.
6
10
-1
1 10
10
2
Neutrino energy in MeV
0
0.2
0.4
0.6
0.8
1

e
s
u
r
v
i
v
a
l
p
r
o
b
a
b
i
l
i
t
y
Vacuum
oscillations
Adiabatic matter
resonance
sin
2

12
1 -
1

2
sin
2
2
12
0 2 4 6 8 10
flux of solar B neutrinos in 10
6
cm
-2
s
-1
0
2
4
6
8
10
f
l
u
x
o
f
s
o
l
a
r
B
e
n
e
u
t
r
i
n
o
s
i
n
1
0
9
c
m
-
2
s
-
1
90, 99% CL (2 dof)
SSM
Figure 4: The energy-dependent survival prob-
ability predicted by LMA, and how experimental
data restrict the low-energy and high-energy lim-
its of P(
e

e
, E

).
Figure 5: Best-t regions at 90, 99% CL (2 dof )
for Boron and Beryllium uxes from a global
solar-model-independent t of solar data. The
horizontal ellipse shows solar model predictions.
In this way the equality in eq. (4) gets replaced by a inequality. In order to access also the other
region we analytically continue to imaginary
13
, where cos
2

13
> 1. (This is analogous but less usual
than allowing sin
2
2
23
> 1 in the atmospheric t). The global t is performed by keeping xed m
2
12
at the central value suggested by KamLAND, because a variation of m
2
12
within the range in table 1
negligibly aects solar data. This is the key extra input provided by KamLAND; a more complicated
solar plus KamLAND global t would give the same result.
The result of the global solar t performed in this specic context is shown in g. 2b, and agrees
with the semi-quantitative general result of eq. (6). We see that the LMA prediction in eq. (4) is well
compatible with data. The constraint on
13
provided by solar data (subdominant with respect to
the constraint from CHOOZ and atmospheric data) is included in the global analysis summarized in
table 1 and g. 1.
We performed more global ts considering a few other ways of avoiding the LMA prediction of
eq. (4): in each case the allowed region looks like a potato similar to the one in g. 2b. Dierent ts
give allowed regions with sizes and shapes that vary roughly as much as potatoes vary. For example we
tried to linearly distort the P(
e

e
, E

) prole predicted by LMA for

13
= 0 as P(
e

e
, E

) =
sin
2

12
+ [P
LMA
(
e

e
, E

) sin
2

12
]. This is somewhat dierent than considering a
13
,= 0,
because matter eects depend on
13
, that therefore does not act as a linear distortion.
The main point is that the approximate general result of eq. (6) fairly summarizes the variety of
exact results obtained by performing global ts in presence of dierent mechanism that distort the
prediction of eq. (4) without introducing new notable features at intermediate energies. Therefore
eq. (4) is a useful semi-quantitative way of summarizing our present knowledge of P(
e

e
). Its
behavior at intermediate energies 1 MeV
<

<

10 MeV is basically unknown.

7
1.4 Low energy neutrinos and BOREXINO
Can future sub-MeV solar neutrino experiments improve on oscillation parameters? This question was
answered in [15], and we do not have much to add; for recent works on the subject see [22]. We recall
here the main points. Low energy solar neutrinos do not experience the MSW resonance in the sun:
their survival probability is therefore given by the averaged vacuum oscillations expression of eq. (7b).
This means that, in rst approximation, sub-MeV experiments have nothing to tell about m
2
12
, but
could give information on
12
by measuring P(
e

e
, small E

). Within the 3 neutrino context

this same survival probability of eq. (7b) is also directly measured by reactor
e
experiments, such as
KamLAND. KamLAND can achieve a determination of P(
e

e
, small E

) competitive with the

solar result in eq. (6). A future reactor experiment with baseline 50km appropriate for observing
the rst oscillation dip could achieve an error (2 3) times lower than the present error in eq. (6).
Such reduced uncertainty is already achieved today, if one trusts the LMA prediction of eq. (4) and
employs it to infer the low-energy limit of P(
e

e
) from SNO data:
P(
e

e
, small E

) = 1
1
2
sin
2
2
12
= 0.57 0.02 (LMA prediction). (8)
For simplicity we here assumed
13
= 0, because its value is not a relevant issue: if
13
is large enough
to have a sizable eect, long-baseline experiments will see and measure it so well that only the central
value of eq. (8) (but not its uncertainty) has to be changed.
The precision of sub-MeV experiments is ultimately limited by the 1% solar model uncertainty
on the pp ux. Although it presently seems unrealistic to aim at measuring the pp rate with this
level of accuracy, the above discussion shows that such ultimate precision should be achieved, if sub-
MeV experiments want to contribute signicantly to the determination of
12
. Going beyond the
3 framework, sub-MeV experiments can make important searches for new physics beyond neutrino
masses. The reason is that high energy solar neutrinos, detected by SNO, are almost pure
2
(the
neutrino mass eigenstate with mass m
2
), so that low energy experiments are needed to probe new
physics that dominantly aects
1
. This argument applies e.g. in presence of an extra sterile neutrino
in wide ranges of its oscillation parameters [23].
If LMA is the end of the story the near-future experiment Borexino will not improve the deter-
mination of oscillation parameters and should observe no anomalous day/night nor seasonal variation.
(KamLAND can also be converted into a solar neutrino experiment). We comment on the impact of
Borexino (and eventually KamLAND) from the point of view of new physics. As discussed below,
in presence of generic new physics, existing data poorly constrain the Beryllium rate, so that large
deviations from the LMA prediction are possible. The rate measured by Borexino can be modied
by new physics that mostly aects intermediate energy neutrino. Rather than performing a detailed
analysis of one specic source of new physics, out purpose will be to study this possibility in an approx-
imate but suciently general way. To this end, we notice that in Borexino the eects of unspecied
new physics dominantly manifest as an anomalous ux of Beryllium neutrinos (Borexino also has a
minor sensitivity to pep neutrinos [24]). Therefore, in the next section we study how well present data
restrict the Beryllium rate.
1.5 Solar neutrino uxes
An eventual deviation of solar neutrino uxes from solar model LMA predictions could be due to new
physics in neutrinos beyond LMA, or to (new) physics not included in solar models. For concreteness
we focus on the second case, and study how well present solar neutrino data determine solar neutrino
uxes, updating the results of [20]. Apparently a detailed global t seems needed to address this issue,
8
but once again one can answer to this question in a simple way. Following [20] the main points
to consider are the following ones:
1. The luminosity constraint allows to precisely predict the pp uxes. Since no experiment so far
is sensitive to deviations compatible with the luminosity constraint, we can essentially set pp
uxes to their solar model value.
2. SNO measured the Boron ux.
3. Only two kind of experiments, Gallium and Chlorine, have measured low-energy neutrino uxes.
Therefore, data only constrain two linear combinations of low-energy uxes.
4. The Chlorine experiment has a poor sensitivity to low energy neutrinos: after subtracting the
80% Boron contribution to the Chlorine rate, as directly measured by SNO via CC, the
residual low-energy contributions to the Chlorine rate is just about 2 above zero.
Therefore the Chlorine rate carries so little information on low energy uxes, that our present knowl-
edge on low-energy uxes is well summarized by a single number: their contribution to the Gallium
rate. Restarting from eq. (5), we now subtract from the total Gallium rate R
Ga
its
8
B contribu-
tion and its pp, pep contributions (as predicted by solar models and LMA oscillations, see eq. (6):
R
Ga
pp,pep
= 41.3 1.5), obtaining
(22.5 4) SNU = R
Ga
CNO
+R
Ga
7
Be
=
4.07
Be
+ 4.6
CNO
10
9
/cm
2
s
SNU. (9)
We have taken into account that LMA oscillations suppress both rates by about 0.550.02 a value
negligibly dierent from the low-energy limit of P(
e

e
) of eq. (8).
In order to show the accuracy of our simplied analysis, we now compare its results with the
exact results of a solar-model-independent global analyses of solar and KamLAND data, performed
as described in [20]. To perform such comparison we need to consider a well dened, simple and
relevant sub-case: we assume a non standard Beryllium ux but a standard CNO ux,
CNO

0.6 10
9
/cm
2
s. As previously discussed, this choice is motivated by the fact that Beryllium neutrinos
are more important, that Borexino should study them, and that according to solar models the CNO
contribution to any measured solar neutrino rates is smaller than its experimental error. In this way
eq. (9) reduces to 7
Be
= (4.9 1.1) 10
9
/cm
2
s. This can be directly compared with the result of the
global analysis, shown in g. 5, that can be approximatively summarized as
8
B
= (5.0 0.2) 10
6
/cm
2
s, (10a)
7
Be
= (4.1 1.1) 10
9
/cm
2
s. (10b)
The global t also includes the Chlorine rate, which has a central value about 2 lower than the
LMA prediction, and consequently somewhat reduces 7
Be
. From these numbers, we see that the
determinations of the Boron and Beryllium uxes that follow from our simplied analysis are quite
adequate. Borexino will signicantly improve over the present determination of 7
Be
.
The simplied analysis also determines our present knowledge of the CNO ux. As previously
discussed solar neutrino experiments basically measured the linear combination of CNO and Beryllium
uxes of eq. (9), that directly implies an upper bound on the CNO ux:

CNO
< 6 10
9
/cm
2
s at 3 (1 dof), (11)
which is one order of magnitude above solar model predictions. Eq. (11) updates the value rst
obtained in [20] (section 4). This constraint was also re-derived in [25], emphasizing that the result of
9
eq. (9) or of eq. (11) proves that the CNO cycle does not give the dominant contribution to the total
solar luminosity L

. Indeed by converting the neutrino ux

CNO
into the corresponding energy ux
L
CNO
, eq. (11) reads L
CNO
<

0.1 L

.
2 Oscillations with atmospheric frequency
Present data do not precisely determine [m
2
23
[ nor
23
(see table 1 or g. 1), give an upper bound on

13
and do not determine the sign of m
2
23
(i.e. if neutrinos have normal or inverted mass hierarchy).
Several experimental programs using long-baseline and reactor neutrinos plan to to conrm the SK
evidence and to improve on m
2
23
,
23
and
13
, possibly measuring a non-zero value of the latter angle.
From the point of view of the 3 neutrino framework these experimental programs seem adequately
complete [26], thanks in particular to JHF [27], possibly complemented with a new reactor neutrino
experiment [28]. Here, we discuss the present determination of atmospheric parameters and other
implications of the existing data.
The global t. The atmospheric t includes the nal SKI data [9], the nal Macro data [10]
and the latest K2K [11] data. As in other similar analyses, the SK and K2K ts are extracted from
the latest SK and K2K papers by graphical reduction (i.e. using a scale and a pencil) because this
procedure guarantees a more accurate result than an independent reanalysis. We employ the standard
analysis of nal SKI data. The SK collaboration also performs an alternative analysis, by selecting
the data which have the best resolution in L/E

, obtaining similar central values and somewhat

dierent uncertainties. (The CPT-violating t of g. 3a is instead based on an independent reanalysis
of atmospheric data, so its CPT-conserving limit can slightly dier from the CPT-conserving t of
g. 1).
The parameter
23
. The value of this parameter can be obtained from a simple physics argument.
It is dominantly determined by SK multi-GeV -like events as sin
2
2
23
2(1N

/N

) = 1.020.08,
where N

400 (N

200) are the number of down-ward (up-ward) going events, that experience
roughly no oscillation (averaged oscillations). A detailed analysis is needed to include the rest of
the SK, K2K, Macro data, which however only mildly improve on the determination of sin
2
2
23
.
An improved measurement of sin
2
2
23
from the up/down ratio of atmospheric neutrinos could be
performed at a future Mton-scale atmospheric detector.
4
Future long-baseline experiments with very
intense conventional neutrino beams [27] will also lead to progress.
The parameter [m
2
23
[. The present situation concerning [m
2
23
[ is quite dierent: SK cannot
precisely measure it and cannot see a clear oscillation dip, and a detailed analysis is necessary to extract
its central value and error. Long-baseline experiments should signicantly reduce the uncertainty
on [m
2
23
[ by identifying the energy at which neutrinos experience the rst oscillation dip. This
measurement has been already performed by K2K, but with poor statistics: K2K achieves [11] a
determination of [m
2
23
[ with central value and error close to the one of SK [9].
The parameter
13
. The CHOOZ constraint on
13
is strongly correlated with the determination
of [m
2
23
[. As discussed in section 1.3 solar data have a subdominant impact on the determination of

13
, comparable to the eect of changing the kind of analysis of SK atmospheric data (we used the
4
One should take into account the few % eects of solar oscillations and of 13 (see [29] for a recent discussion) which
will be respectively better measured by solar and reactor/long-baseline experiments.
10
10
-3
0.01 0.1 1
lightest neutrino mass in eV
0.1
1
0.03
0.3
3
m
c
o
s
m
o

i
n
e
V
99% CL (1 dof)
disfavoured by cosmology
m
23
2
< 0
m
23
2
> 0
10
-3
0.01 0.1 1
lightest neutrino mass in eV
0.01
0.1
1
0.003
0.03
0.3
3
m

e

i
n
e
V
99% CL (1 dof)
Bound from MAINZ and TROITSK
Sensitivity of KATRIN
d
i
s
f
a
v
o
u
r
e
d
b
y
c
o
s
m
o
l
o
g
y
m
23
2
< 0
m
23
2
> 0
Figure 6: 99% CL expected ranges for the parameters m
cosmo
= m
1
+ m
2
+ m
3
probed by cosmology
(g. 6a) and m
e
(mm

)
1/2
ee
probed by -decay (g. 6b) as function of the lightest neutrino mass. The
darker lines show how the ranges would shrink if the present best-t values of oscillation parameters
were conrmed with negligible error.
standard SK analysis). The statistically insignicant hint for a
13
> 0 in g. 1 is mainly due to a
small decit of events in CHOOZ data at lowest energies.
Other eects? Data show no signicant hint for new eects beyond three massive neutrinos. For
example g. 3a shows a global t performed without assuming that neutrinos and anti-neutrinos
have the same atmospheric mass splitting and mixing angle. We see that the best-t lies close to
the CPT-conserving limit, and that the atmospheric mass splitting in anti-neutrinos is poorly deter-
mined. Nevertheless, this is enough to strongly disfavor a CPT-violating interpretation of the LSND
anomaly [19]. Near-future long-baseline experiments will probably study only rather than .
3 Non-oscillation experiments
In this section we discuss non-oscillation experiments and consider the 3 non-oscillation parameters
mentioned in the introduction. Making reference to experimental sensitivities, the 3 probes should
be ordered as follows: cosmology, 02 and nally decay. Ordering them according to reliability
would presumably result into the reverse list: cosmological results are based on untested assumptions,
and 02 suers from severe uncertainties in the nuclear matrix elements. Even more, there is an
interesting claim that the 02 transition has been detected [12] (see section 3.3 for some remarks),
there is a persisting anomaly in TROITSK decay, and even in cosmology, there is one (weak) claim
for a positive eect. None of these hints can be considered as a discovery of neutrino masses. Several
existing or planned experiments will lead to progress in a few years.
In this section, we assume three massive Majorana neutrinos and study the ranges of neutrino
mass signals expected on the basis of oscillation data, updating and extending the results of [30].
Our inferences are summarized in table 1 and obtained by marginalizing the full joint probability
distribution for the oscillation parameters, using the latest results discussed in the previous sections.
11
3.1 Cosmology
Cosmological observations are mostly sensitive to the sum of neutrino masses: m
cosmo
= m
1
+m
2
+m
3
,
that according to standard cosmology controls the energy fraction

in non relativistic neutrinos as

h
2
= m
cosmo
/93.5eV where as usual h parameterizes the present value of the Hubble constant as
h H
today
/(100km/s Mpc). Cosmology does not distinguish Majorana from Dirac neutrino masses.
In order to convert CMB and LSS data into a constraint on neutrino masses one needs to assume a
cosmological model. The WMAP constraint [31] assumes that the observed structures are generated
by Gaussian adiabatic primordial scalar uctuations with at spectral index n evolved in presence of
the known SM particles, of cold dark matter and of a cosmological constant. This standard model of
cosmology seems consistent with all observations. Furthermore, it is assumed that observed luminous
matter tracks the dark matter density up to a scale-independent bias factor. The data-set includes
Lyman- data about structures on scales so small that perturbations are no longer a minor correction
to a uniform background. These data are sensitive to neutrino masses (and thereby somewhat aect
the global cosmological t) but might be aected by non linear evolution eects, which are not fully
understood. In summary, cosmology presently gives the dominant constraint, which however rests on
untested assumptions and on risky systematics.
In the future the sensitivity to neutrino oscillations will improve thanks to better CMB data and
to new LSS measurements less plagued by potential systematic eects. If cosmology were simple (e.g.
a spectral index n = 1, no tensor uctuations,. . . ) then it seems possible to detect even neutrino
masses as small as allowed by oscillation data [32]. The expected ranges of m
cosmo
are reported in the
lowest row of table 1 in the limiting case where the lightest neutrino is massless, and in g. 6a in the
general case.
3.2 Direct search via decay
-decay experiments dominantly probe the quantity m
e
(m m

)
1/2
ee
. If neutrinos are quasi-
degenerate, m
e
is their common mass. The MAINZ and TROITSK experiments obtained comparable
limits: m
2
e
= 1.22.22.1 eV
2
[33] and m
2
e
= 2.32.52.0 eV
2
[34] respectively. The 95% bounds
quoted by the experimental collaborations agree with the values obtained in Gaussian approximation.
Thus, we combine the two measurements by summing errors in quadrature and get
m
2
e
= 1.7 2.2 eV
2
i.e. m
e
< 2.0 eV at 99 % C.L. (12)
In order to study what oscillation data imply on the value of m
e
we write it in terms of the neutrino
masses m
i
and of the mixing angles
ij
as
m
e
=

cos
2

13
(m
2
1
cos
2

12
+m
2
2
sin
2

12
) +m
2
3
sin
2

13

1/2
. (13)
In the case of normal mass hierarchy, m
1
m
2
m
3
oscillation data imply the 99% CL range
m
e
= (4.610) meV. In the case of inverted hierarchy, m
3
m
1
< m
2
one gets m
e
= (4257) meV.
The last number if a factor 5 below the planned sensitivity of Katrin [35].
It is immediate to obtain the ranges corresponding to the generic case of a non vanishing lightest
neutrino mass m
lightest
(where m
lightest
= m
1
in the case of normal hierarchy and m
lightest
= m
3
in the
case of inverted hierarchy). As clear from the denition m
2
e
(m m

)
ee
or from the more explicit
expression in eq. (13) one just needs to add m
2
lightest
to m
2
e
. The resulting bands at 99% C.L. are
plotted in g. 6b.
12
Nucleus and observed background expected 99% C.L. bound
experiment events, n events, b signal on [m
ee
[/h
76
Ge HM [38] 21 20.4 1.6 76 [m
ee
/eV[
2
/h
2
0.44 eV
76
Ge IGEX [36, 44] 9.6 17.2 2 23.5[m
ee
/eV[
2
/h
2
0.55 eV
130
Te Cuoricino [45, 37] 24 35.2 4 21.5[m
ee
/eV[
2
/h
2
0.62 eV
Table 2: Numbers of observed events, expected background and predicted signal in the most sensi-
tive 02 experiments. The last column shows the 99% C.L. constraint on [m
ee
[/h, where h 1
parameterizes the uncertain 02 nuclear matrix element and depends on the nucleus studied.
3.3 Neutrino-less double-beta decay
Updating the results of [30], in the 3 neutrino framework we discuss the connection with oscillations,
the bound from 02 on neutrino masses and the possible hint of a signal.
IGEX, CUORICINO and nuclear uncertainties. First, we recall that recently two experiments
produced new relevant data. The rst is IGEX [36], now terminated, which used 86% enriched
76
Ge
(with exposure 7 10
25
nucleiyr). The second is Cuoricino [37], now running, which uses 34%
natural
130
Te (with exposure 1.4 10
24
nucleiyr, and the plan to collect 10 times more data in about 6
years). Both experiments reached a good level of background (about 0.18/keV kg yr), good detection
eciencies of 70% and 84%, and respectable energy resolutions, the full width at half maximum
being 4 keV and 7 keV respectively. For comparison, the HeidelbergMoscow (HM) data-set with
pulse-shape-discrimination [38] has exposure 2.5 10
26
nucleiyr and background 0.06/(keV kg yr).
As in [30] we quote the limits on [m
ee
[ adopting the 02 nuclear matrix elements /
0
computed
in [39]. To use a dierent calculation with matrix element /, just rescale by the factor h = /
0
//,
which depends on the nucleus studied.
5
We always explicit the factors h when quoting an experimental
result on 02. We prefer to show such uncertainty explicitly rather than attempting to evaluate the
theoretical error on matrix elements.
6
Dierent published computations nd h in the following ranges
0.3 [40] < h(
76
Ge) < 2.4 [41], 0.4 [40] < h(
130
Te) < 2.7 [42]. (14)
When comparing two results on [m
ee
[ obtained with dierent nuclei one needs to consider the ratios
of h, e.g., h(
130
Te)/h(
76
Ge). This quantity is also uncertain, spanning the following range:
0.3 [43] < h(
130
Te)/h(
76
Ge) < 1.7 [42]. (15)
Now we convert 02 data into a constraint on [m
ee
[ following the same simple procedure employed
by the HM collaboration [38]: the Poisson likelihood of having s signal events is L(s) e
s
(b + s)
n
,
where n and b are the numbers of observed and expected background events in the 3
E
window around
the Q value of the 02, and
E
is the energy resolution of the apparatus. We therefore evaluate n
and b in the 10 keV region around Q = 2039 keV for IGEX [36, 44] and in the 18 keV region around
Q = 2529 keV for Cuoricino [45, 37]. Results are collected in table 2. For both experiments the
observed number of events is slightly below the expected background. In this paper we systematically
employ the Gaussian approximation: the 99% constraint (2.58) on [m
ee
[ is computed by dening

2
2 ln L, marginalizing it with respect to the uncertainty in the background, and nally quoting
the value of [m
ee
[ such that
2
(m
ee
)
2
(m
ee
= 0) = 2.58
2
. Dierent approaches to statistical
inference give slightly dierent constraints.
5
The symbol h has dierent meaning for cosmology and for 02, as should be clear from the context.
6
It would be useful if future calculations of matrix elements could provide error estimates.
13
Since HM and IGEX are both based on
76
Ge, it is possible to obtain a more stringent combined
bound, [m
ee
[ < 0.38 h eV.
7
Inference on [m
ee
[ from oscillations. Using the latest oscillation data, we study the expected
range of [m
ee
[. We follow the procedure described in [30]. Assuming three CPT-invariant massive
Majorana neutrinos we get:
a) In the case of normal hierarchy (i.e. m
1
m
2
m
3
, or m
2
23
> 0) the ee element
of the neutrino mass matrix probed by 02 decay experiments can be written as [m
ee
[ =
[e
2i
m
sun
ee
+ e
2i
m
atm
ee
[ where , are unknown Majorana phases and the solar and atmo-
spheric contributions can be predicted from oscillation data. The solar contribution to m
ee
is m
sun
ee
= (m
2
12
)
1/2
sin
2

12
cos
2

13
= (2.78 0.22) meV. The bound on
13
implies m
atm
ee
=
(m
2
23
)
1/2
sin
2

13
< 1.7 meV at 99% CL. By combining these two contributions we get the range
reported in table 1. The precise prediction is shown in g. 7a. Unlike in our previous results [30],
a non zero contribution is now guaranteed at high condence level, because data now tell that
the solar contribution is larger than the atmospheric contribution, so that a cancellation is
not possible.
b) In the case of inverted hierarchy (i.e. m
3
m
1
m
2
, or m
2
23
< 0), from the prediction
[m
ee
[ [m
2
23
[
1/2
[ cos
2

12
+ e
2i
sin
2

12
[ cos
2

13
we get the range reported in table 1 (see
also [46]). The precise prediction is shown in g. 7b. The main uncertainty on [m
ee
[ is due to
the Majorana phase , rather than to the oscillation parameters. If the present central values
were conrmed with innite precision we would still have the loose range [m
ee
[ = (19 50) meV
at any C.L.
The [m
ee
[ ranges for normal and inverted hierarchy do not overlap. Values of [m
ee
[ outside these
ranges are possible if the lightest neutrino mass is not negligible, as shown in g. 7d. In the inverted
hierarchy case (m
2
23
< 0) a non zero lightest neutrino mass can only make [m
ee
[ larger than in case
b). The darker regions in g. 7d show how the predicted range of [m
ee
[ would shrink if the present
best-t values of oscillation parameters were conrmed with innite precision. The two funnels present
for m
2
23
> 0 have an innitesimal width because we assume
13
= 0 and would have a nite width
if
13
,= 0.
c) The case of quasi-degenerate neutrinos with mass m
e
corresponds to the upper region of
g. 7d. [m
ee
[ and m
cosmo
are related to m
e
as
m
cosmo
= 3m
e
and 0.24 m
e
< [m
ee
[ < m
e
at 99% C.L. (16)
The lower bound on [m
ee
[ holds thanks to the fact that solar data exclude a maximal solar
mixing and that CHOOZ requires a small
13
.
Upper bound on neutrino masses from 02. The above result means that by combining
oscillation data with the 02 upper bound on [m
ee
[ implies an upper bound on the parameter m
e
probed by decay experiments [30]. In view of present values, such m
e
corresponds to the common
mass of quasi-degenerate neutrinos. This bound is shown in g. 7c, and depends on the 02 nuclear
matrix elements, parameterized by the uncertain factors h 1. Therefore the results of Cuoricino,
obtained with a nucleus dierent than HM and IGEX, add condence in the result. For h = 1 the
7
We cannot combine HM/IGEX with Cuoricino in a reliable manner, due to the uncertainty on the relative nuclear
matrix element discussed above. Using the matrix elements of [39] for
76
Ge and
130
Te, would give |mee| < 0.34 h eV at
99% C.L. with h = 1.
14
0 1 2 3 4 5 6
| m
ee
| in meV
0
2
4
6
8
10
12
14

2
Normal hierarchy
90 %CL
99 %CL
0 10 20 30 40 50 60 70
| m
ee
| in meV
0
2
4
6
8
10
12
14

2
Inverted hierarchy
90 %CL
99 %CL
0 0.5 1 1.5 2
m

e
/ h in eV
0
2
4
6
8
10
12
14

2
Degenerate neutrinos
90 %CL
99 %CL
C
u
o
r
i
c
i
n
o
H
M
I
G
E
X
10
-4
10
-3
10
-2
10
-1
1
lightest neutrino mass in eV
10
-4
10
-3
10
-2
10
-1
1
|
m
e
e
|

i
n
e
V
99% CL (1 dof)
m
23
2
> 0
disfavoured by 02
d
i
s
f
a
v
o
u
r
e
d
b
y
c
o
s
m
o
l
o
g
y
m
23
2
< 0
Figure 7: Predictions for [m
ee
[ assuming a hierarchical (g. 7a) and inverted (g. 7b) neutrino spec-
trum. In g. 7c we update the upper bound on the mass of quasi-degenerate neutrinos implied by 02
searches. The factor h 1 parameterizes the uncertainty in the nuclear matrix element (see sect.
2.1). In g. 7d we plot the 99% CL range for m
ee
as function of the lightest neutrino mass, thereby
covering all spectra. The darker regions show how the m
ee
range would shrink if the present best-t
values of oscillation parameters were conrmed with negligible error.
15
2000 2020 2040 2060 2080 2100
Energy in electrons in keV
0
5
10
15
20
25
c
o
u
n
t
s
8 9 10 11 12 13
background/keV
0
1
2
3
4
5
e
v
i
d
e
n
c
e
f
o
r
0

i
n

Figure 8: Fig. 8a: the latest HM data [13] (71.7kg yr) used to claim a 4.2 evidence for 02.
Fig. 8b: the statistical signicance of the 02 signal, as function of the assumed at component of
the background.
combined constraint is m
e
/h < 1.0 eV at 99% C.L. This constraint is stronger than the -decay
constraint (but holds under the additional assumption that neutrinos are Majorana particles) and is
weaker than the cosmological constraint (but needs no assumptions about cosmology).
In all above discussions we employed the bound on [m
ee
[ from HM data, as published by the HM
collaboration [38].
Remarks on the hint for 02. While the HM collaboration used their data to set a bound on
[m
ee
[, some members of the HM collaboration reinterpreted the data as an evidence for 02 [12].
According to this claim, the latest HM data [13] plotted in g. 8a contain a 4.2 evidence for a 02
peak (indicated by the arrow). In these latest results the peak is more visible than in latest published
HM data [38], partly thanks to higher statistics (increased from 53.9 to 71.7 kg yr) and partly thanks
to an improved analysis [13]).
This claim is controversial, mainly because one needs to fully understand the background before
being condent that a signal has been seen. In order to allow a better focus on this key issue, we
present g. 8b, which should be uncontroversial. It shows the statistical signicance
8
of the 02
signal as a function of the true background level b, assumed to be quasi-at close to the Q-value of
02, Q 2039 keV. The crucial issue is: how large is b? The HM collaboration earlier claimed [38]
b = (13.6 0.7)events/(71.7 kg yr keV). In such a case the statistical signicance of the signal would
be less than 1, see g. 8b. This can be considered as the upper bound on b computed assuming that
all events in a wider range around Q come from a quasi-at background.
A statistically signicant hint for 02 is obtained if one can show that b is lower. The continuous
line in g. 8a shows a t of HM data using a tentative model of the background [12], assumed to have
a quasi-at component b (mainly due to natural and cosmogenic radioactivity) plus some peaks due
to faint lines of
214
Bi, which is a radioactive impurity present in the apparatus (from the
238
U decay
chain). Their positions and intensities can be estimated from tables of nuclear decays; however they
are modied by O(1) factors by detector-related eects which depend on the unknown localization of
214
Bi (see [13] and ref.s therein). The t in g. 8a is performed by allowing the intensity of each line
to freely vary. In this way part of the background is interpreted as
214
Bi peaks, thereby reducing the
8
Dened in Gaussian approximation as (
2
no signal

2
best t
)
1/2
where
2
= 2 lnL and L is the likelihood computed
combining statistical Poissonian uncertainties with other systematic uncertainties.
16
quasi-at component b. Proceeding in this way, we nd that the statistical signicance of the 02
signal is about 2.7. This is the best we can do with the data at our disposal. This kind of analysis
was proposed in [30] and has been adopted in [13].
However, some details in its implementation prevent this analysis from fully reaching its goal,
which is determining b from regions with no peaks. 1) The latest data have been published only
below 2060 keV. (Above 2060 keV in g. 8a we plotted HM data articially rescaled to account for
the larger statistics). Below 2060 keV there are little energy ranges with no peaks. Including the old
data above 2060 keV in the t would reduce the signicance of the signal down to about 2.2. 2) HM
data contain hints of extra unidentied spurious peaks at specic energies (at 2030 keV and above
2060 keV). Fitting data assuming that these extra peaks can be present at arbitrary energies with
arbitrary intensities reduces b and enhances the statistical signicance of the signal.
Various future experiments plan to test the claim of [12]. The most direct test requires using the
same technique (germanium detectors) but in the worst case, when the observed hint of a signal is
due to some irreducible (hypothetical) background, a safe test requires a dierent type of detector.
As previously discussed Cuoricino employs a dierent nucleus: its capabilities relative to HM and
IGEX depend on the uncertain relative nuclear matrix element. For instance, with the lowest value in
eq. (15) Cuoricino is already testing the claim of [12]; for intermediate values new data of Cuoricino
will signicantly test the claim; for the highest value Cuoricino will be not sucient.
In our view, a discussion of what should be considered as a convincing evidence for 02, is anyway
useful or necessary, because any experiment (past and future) needs to confront with this issue.
4 Conclusions
Assuming oscillations of three active massive neutrinos we updated the determination of the oscillations
parameters, at the light of latest experimental data. Results are shown in table 1 and in g. 1. We
notice that the parameters m
2
12
,
12
,
23
are now dominantly determined by simple and robust sub-
sets of data, such that simple arguments give the same nal result as global analyses. Pieces of data
that play a sub-dominant r ole in parameter determination allow to test our assumptions: e.g. allowing
neutrinos and anti-neutrinos to have dierent masses and mixings gives the CPT-violating t of g. 3.
Present data do not contain evidence for extra eects.
We also tried to study in a simple and general way related topics, such as the determination of
the limiting survival probability of solar neutrinos at small and large energies, the present knowledge
of solar neutrino uxes, etc. Approximated general results have been compared with exact results of
global ts performed in specic cases.
Finally, updating the results of [30], we studied how oscillation data allow to infer the combination
of neutrino masses probed by cosmology, -decay and 02-decay experiments, and discussed the
present experimental situation.
References
[1] The results of the Homestake experiment are reported
in B.T. Cleveland et al., Astrophys. J. 496 (1998) 505.
[2] Latest Gallex and SAGE data have been pre-
sented in a talk by C. Cattadori at the Neutrino
2004 conference (Paris, 14-19 June), web site neu-
trino2004.in2p3.fr.
[3] Gallex collaboration, Phys. Lett. B447 (1999) 127.
[4] SAGE Collaboration, J. Exp. Theor. Phys. 95 (2002)
181 [astro-ph/0204245].
[5] Super-Kamiokande collaboration, hep-ex/0205075.
[6] SNO collaboration, nucl-ex/0204008 and
nucl-ex/0204009.
[7] SNO collaboration, nucl-ex/0502021.
[8] KamLAND collaboration, hep-ex/0406035.
[9] SuperKamiokande collaboration, hep-ex/0501064.
[10] MACRO collaboration, Phys. Lett. B566 (2003) 35
[hep-ex/0304037].
[11] K2K collaboration, hep-ex/0411038.
17
[12] H. V. Klapdor-Kleingrothaus, A. Dietz, H. L. Harney
and I. V. Krivosheina, Mod. Phys. Lett. A 16 (2001)
2409 [hep-ph/0201231].
[13] H. V. Klapdor-Kleingrothaus, A. Dietz,
I. V. Krivosheina and O. Chkvorets, Nucl. Instrum.
Meth. A 522 (2004) 371 [hep-ph/0403018].
[14] LSND collaboration, Phys. Rev. D64 (2001) 112007
[hep-ex/0104049].
[15] A. Strumia and F. Vissani, JHEP 0111 (2001) 048
[hep-ph/0109172].
[16] J.N. Bahcall, S. Basu, M.H. Pinsonneault, Astrophys.
J. 555 (2001) 990 [astro-ph/0010346].
[17] LUNA collaboration, nucl-ex/0312015.
[18] P. Creminelli et al., J. HEP 05 (052) 2001
[hep-ph/0102234]. Its e-print version, has been up-
dated including the recent data from SNO.
[19] A. Strumia, hep-ph/0201134. The solar CPT-violating
t has been re-emphasized in H. Murayama, Phys.
Lett. B 597 (2004) 73 [hep-ph/0307127] and in A.
de Gouvea, C. Pe na-Garay, hep-ph/0406301. The at-
mospheric CPT-violating t has been also studied by
the SK collaboration, see the talk by E. Kearns at
the Neutrino 2004 conference (Paris, 14-19 June), web
site neutrino2004.in2p3.fr. See also M. C. Gonzalez-
Garcia, M. Maltoni and T. Schwetz, Phys. Rev. D 68
(2003) 053007.
[20] A. Strumia and R. Barbieri, JHEP 12 (2000) 016. Its
e-print version, hep-ph/0011307, contains updates and
additional discussions not present in the published ver-
sion.
[21] S. Goswami, A. Bandyopadhyay and S. Choubey,
Nucl. Phys. Proc. Suppl. 143 (2005) 121
[hep-ph/0409224]. B. S. Koranga, M. Narayan
and S. Uma Sankar, hep-ph/0503092.
[22] J.N. Bahcall, C. Pe na-Garay, JHEP 004 (2003) 0311.
A. Bandyopadhyay, S. Choubey, S. Goswami, S.T.
Petcov, hep-ph/0410283.
[23] For a recent extensive study see M. Cirelli et al,
Nucl. Phys. B708 (2005) 215 [hep-ph/0403158]. For
short summaries see A. Strumia, Nucl. Phys. Proc.
Suppl. 143 (2005) 144 [hep-ph/0407132] and M. Cirelli,
astro-ph/0410122.
[24] C. Galbiati, A. Pocar, D. Franco, A. Ianni, L. Cado-
nati, S. Schonert, hep-ph/0411002.
[25] J. N. Bahcall, M. C. Gonzalez-Garcia and C. Pena-
Garay, Phys. Rev. Lett. 90 (2003) 131301.
[26] For a recent summary of capabilities of neutrino beam
experiments see M. Lindner, hep-ph/0503101. If 13 =
0 it seems practically impossible to discriminate the
kind of neutrino mass hierarchy with oscillation exper-
iments, as recently discussed in A. de Gouvea, J. Jenk-
ins and B. Kayser, hep-ph/0503079.
[27] Y. Itow et al., The JHF-Kamioka neutrino project,
hep-ex/0106019
[28] P. Huber, M. Lindner, T. Schwetz and W. Winter,
Nucl. Phys. B 665 (2003) 487 [hep-ph/0303232].
[29] O.L.G. Peres, A.Yu. Smirnov, Nucl. Phys. B680 (2004)
479 [hep-ph/0309312].
[30] F. Feruglio, A. Strumia and F. Vissani, Nucl. Phys.
B 637 (2002) 345 [addendum-ibid. B 659 (2003) 359]
[hep-ph/0201291].
[31] C.L. Bennett et al., astro-ph/0302207 and D.N.
Spergel et al., astro-ph/0302209. The WMAP con-
straint is similar to previous analyses, e.g. A. Lewis,
S. Bridle, Phys. Rev. D66 (2002) 103511. Other
(sometimes more conservative) analysis nd similar
(sometimes weaker) bounds: S. Hannestad, JCAP
0305 (2003) 004 [astro-ph/0303076], S.W. Allen, R.W.
Schmidt, S.L. Bridle, Mon. Not. Roy. Astron. Soc.
346 (2003) 593 [astro-ph/0306386], M. Tegmark et
al. [SDSS Collaboration], Phys. Rev. D69 (2004)
103501 [astro-ph/0310723] V.Barger et al., Phys. Lett.
B595 (2004) 55 [hep-ph/0312065], P. Crotty, J. Les-
gourgues, S. Pastor, Phys. Rev. D69 (2004) 123007
[hep-ph/0402049], U. Seljak et al., astro-ph/0407372.
[32] For recent discussions see S. Hannestad, Phys. Rev.
D 67 (2003) 085017 [astro-ph/0211106]; J. Lesgour-
gues, S. Pastor and L. Perotto, Phys. Rev. D 70 (2004)
045016 [hep-ph/0403296].
[33] MAINZ collaboration, hep-ex/0412056.
[34] TROITSK collaboration, Phys. Lett. B 460 (1999)
227. The latest results have been presented in a talk
by V. Lobashev at the XI International Workshop on
Neutrino Telescopes (Venezia, Feb. 22-25 2005).
[35] KATRIN web site: www-ik1.fzk.de/tritium.
[36] IGEX collaboration, Phys. Rev. C 59 (1999) 2108.
[37] Cuoricino collaboration, hep-ex/0501034.
[38] Heidelberg-Moscow collaboration, Eur. Phys. J. A 12
(2001) 147 [hep-ph/0103062].
[39] A. Staudt et al., Europh. Lett 13 (1990) 31.
[40] A. Staudt et al., Phys. Rev. C46 (1992) 871.
[41] E. Caurier, Phys. Rev. Lett. 77 (1996) 1954.
[42] V.A. Rodin et al., Phys. Rev. C68 (2003) 044302.
[43] J. Engel et al., Phys. Rev. C37 (1988) 871.
[44] C. E. Aalseth et al., Mod. Phys. Lett. A 17 (2002)
1475 [hep-ex/0202018].
[45] Cuoricino collaboration, Phys. Lett. B 584 (2004)
260.
[46] H. Murayama and C. Pena-Garay, Phys. Rev. D 69
(2004) 031301 emphasized that the lower bound on
|mee| in the case of inverted hierarchy is robust and
found a value in agreement with results in [30].
18