In Aussois, the session on ``Present and Future Neutrino Beams'' reviewed the long baseline experiments that will help to understand the neutrino mixing parameters phenomenology in the coming years. This is a long and step-by-step process. In the first step, the MINOS (see M. Bishai's talk), OPERA and ICARUS (see D. Duchesneau's talk) will confirm and improve the SuperK atmospheric oscillation result. This phase will provide an improvement of the limit on \thetachooz ($\simeq < 0.06$ 90\% C.L.) In the second step, T2K (see Kobayashi's talk) and NO$\nu$A (see Ray's talk) will focus on measuring \thetachooz ($\simeq < 0.006$ 90\% C.L.) . This measurement is a prerequisite before attempting to look for CP violation in the leptonic sector: this will be the task of the third step, and the VLBL (see M. Bishai's talk), the T2K-II (see Kobayashi's talk) and the CERN-Fr\'ejus (see M. Mezzetto's talk) proposals. The ultimate tool in neutrino physics -- the neutrino factory -- was not discussed in this meeting. \section{First step} There are presently four experiments running or planned to confirm the atmospheric neutrino result and improve on the knowledge of the oscillation parameters (\deltaatm, \sinatm): K2K, MINOS, OPERA and ICARUS. The last 3 will also search for the sub-leading \numunue ~oscillations, attempting at a first measurement of the \thetachooz angle. The four experiments rely on very different experimental options (beam and/or detector techniques). \subsection{K2K} (http://neutrino.kek.jp/) The K2K (KEK to Kamioka) long-baseline neutrino oscillation experiment, is the first accelerator-based project to explore neutrino oscillations in the same \deltaatm\ region as the atmospheric neutrinos. By using a low energy \numu\ beam and a flight distance of 250~km, the oscillation process should manifest itself as a reduction of the \numu\ flux at Kamioka (a disappearance) since the \nutau\ produced in the oscillation are below the CC threshold. In addition, the energy spectrum of the observed \numu\ should also be affected by the oscillation. The K2K neutrino beam is produced by 12~GeV protons from the KEK proton synchrotron. The positively charged secondary particles, mainly pions, are then focused by a horn system. The resulting neutrino beam is $98$~\% pure $\nu_\mu$ with a mean energy of $1.3$~GeV. It traverses first the near detector (ND) system, located 300~m downstream from the proton target, and then the SuperKamioka detector, 250~km away. To estimate the ratio of neutrino flux and spectra between Kamioka and KEK (far to near, F/N), a combination of experimental measurements and simulation has to be done. Indeed, due to different geometrical acceptances (the neutrino production place cannot be approximated to a point as seen from the near detector), the neutrino spectra seen by the 2 detectors differ, even with no oscillation. The beam MC simulation is first tuned on the PIMON measurements (pion monitoring detectors intermitently installed upstream of the decay pipe) and then used to compute the F/N ratio in energy bins. This ratio allows then to extrapolate to SK the integrated flux and energy spectrum as measured in the ND, before neutrino oscillate. %(1KT Water Cherenkov detector , %SciFi and SciBar detectors). This extrapolation is compared to the SuperKamiokande measurements. The latest results from the K2K experiment quotes $151^{+12}_{-10}$ fully contained events expected in SK and 107 observed. The main sources of uncertainty in this experiment are the following: \begin{itemize} \item The Monte Carlo F/N ratio relies on a neutrino interaction model, which includes QE, single meson production via baryon resonance and coherent pion production. The relative importance of these componants as a function of energy is poorly known. \item The efficiency of the 1KT Cherenkov detector (affecting the normalization) is dominated by uncertainties in the fiducial volume. \item The energy scale in both Cherenkov detectors %\item autres?? \end{itemize} A two flavor neutrino oscillation analysis for $\nu_\mu$ disappearance is performed by the maximum-likelihood method, using both the number of events and the spectrum shape. The best fit point in the physical region is found at (\sinatm, \deltaatm)=$(1.0, 2.8\times 10^{-3}~{\rm eV^2})$. This result is consistent with the results from atmospheric neutrinos. The final result of the experiment is expected by the end of 2005. \subsection{MINOS} (http://www-numi.fnal.gov/Minos/) MINOS is a long baseline neutrino oscillation experiment utilizing the NuMI beam at Fermilab. This beam is obtained through an intense (0.25 MW) proton beam hitting a graphite target at 120 GeV/c. The movable 2 horn focussing system allows for selecting different energy spectra: low, medium and high. The experimental setup consists of 2 detectors separated by 730 km and as identical as possible: same transverse and longitudinal granularity, same composition and modularity. The basic components are magnetized iron plates interlayed with scintillator strips (the good timing resolution allows background rejection from atmospheric neutrinos and separation of events piling up in the near detector) The main aim of the experiment is to probe the region of parameter space indicated by the atmospheric neutrinos, to demonstrate the oscillation behaviour and to make a precise measurement of \deltaatm\ in a high statistics beam experiment. This will be done through \numu\ disappearance: by plotting, as a function of energy, the ratio of the yield at the far detector to the one expected from near detector measurements. The location and depth of the dip will allow to measure \deltaatm\ and \sinatm . The low energy beam spectrum is best suited to match the latest SK results and has been choosen as the present running condition. A measurement at 10\% (3 years at upgraded intensity: $25\times 10^{20} pot$) %??? can be achieved and could then also rule out exotic oscillation models. The limitation in sensitivity comes mainly from statistics and from the uncertainty in the extrapolation process of the neutrino spectrum from the near to the far detector. A second goal of the experiment is the search for the sub-dominant \numunue\ oscillations aiming at a first \thetachooz\ measurement. In this \nue\ appearance search, the background is dominated by NC produced \pizero 's and some intrinsic \nue\ from the beam. But the sensitivity could reach $\sinchooz < 0.07 $ (twice better as the CHOOZ limit). The detector has been extensively calibrated, both at Cern with a micro-MINOS and with cosmic muons (the shadow of the moon has been observed!). The NuMI beam line has been successfully commissioned and the near and far detectors are presently fully operational with more than 90\% live time. They both have observed their first beam neutrinos in March this year. \subsection{OPERA} (http://operaweb.web.cern.ch/operaweb/) The main goal of OPERA is to focus on providing an unambiguous evidence for \numunutau ~oscillations in the region of the oscillation parameters indicated by the atmospheric neutrino results by looking for \nutau ~appearance in a \numu ~beam. This implies both a high energy beam (above the $\tau $ production threshold) and a very precise tracking detector to observe the $\tau $ produced in charged current \nutau\ interactions (through the characteristic short kink of the $\tau $ decay). The CNGS \numu\ beam will have an average energy of about 17 GeV, a \nue\ and \nuebar\ contamination of 0.87\% and a neglegible prompt \nutau\ contamination. This high energy choice means that, given the 730 km distance between the neutrino source (at CERN) and the OPERA detector, the experiment will be running off the oscillation peak (as expected from the atmospheric neutrino results) and thus will not be sensitive to the oscillation pattern. The $\tau$ detection relies on the photographic emulsion technique. 200000 bricks (emulsion and lead sanwiches) amounting to 1.8 kt will be complemented by electronic trackers and a muon spectrometer.. Although the expected signal is very low (115 \nutau\ CC interactions in the detector and 13 identified by the selection procedure) the expected background will stay around 1 event: the evidence for \numunutau\ should then be very clear. Besides the technical difficulties and complexity of constructing the apparatus, the chalenges in this experiment mainly concern efficiencies: tracking, matching between tracker and emulsions, scanning and selection. Given the good electron identification capabilities in the emulsion bricks, OPERA can look as well for \numunue\ oscillations. The main backgrounds to the \numunue ~oscillations search are the \pizero\ identified as electrons in \numu\ neutral current events, the intrinsic \nue\ beam contamination and the electrons coming from $\tau$ decays in \numunutau\ oscillations. The signal to background ratio can however be enhanced by performing a simultaneous fit to the distribution of the visible energy, electron energy and missing transverse momentum. This yields a 5 years sensitivity corresponding to an upper limit on $sin^2(2 \theta_{13})$ of 0.06 for the nominal beam, similar to the MINOS sensitivity. \subsection{ICARUS} (http://www.aquila.infn.it/icarus/) Located in the Gran Sasso Laboratory in the same CNGS beam as OPERA, the ICARUS experiment is a liquid Argon TPC with imaging capabilities, able to produce high granularity 3D reconstruction of recorded events. %as well as high precision measurements over large sensitive volumes. The operating principle of the LAr TPC is based on the fact that in highly purified LAr ionization tracks can indeed be transported undistorted by a uniform electric field over distances of the order of meters. The detector is not only a tracking device with a precise event topology reconstruction but it can also estimate momentum via multiple scattering, measure local energy deposition ($dE/dx$, providing $e/\pi^0$ separation and particle identification via range versus $dE/dx$ measurement) and reconstruct the total energy of the event from charge integration providing excellent accuracy for contained events. A 600 ton prototype has been extensively tested at surface during the summer 2001, demonstrating that the LAr TPC technique can be operated at the {\it kton} scale with a drift length up to 1.5 {\it m}. Installation at the Gran Sasso Underground Laboratory is currently on-going. Cloning the T600 module will permit to gradually increase the mass and reach a sensitive mass of 2.35ktons. In the present sensitivity estimates only 3 such modules are used since this is the actual guaranteed funding level. Given the good detector performance and the beam conditions (energy and flux), ICARUS will also % as OPERA and MINOS have as a first goal to prove the \numunutau\ oscillation. The analysis will rely on the golden channel: $\tau \rightarrow e\nue\nutau$. The suppression of the background, dominated by \nue\ CC interactions from the beam contamination, will be done through a kinematical analysis, using a 3-dimensional likelihood, including visible energy and missing transverse momentum. With a T1800 detector and 5 years data taking (2.25 $10^{20}$ pot) 6.2 signal events at $2.5 \times 10^{-3} eV^2$ ($\epsilon . BR = 6\%$) for 0.3 background event are expected. The excellent capability of identifying electrons by the ICARUS detector obviously allows to also search for \nue\ appearance. This \numunue\ oscillation component would appear as a distorsion in the energy spectrum of the \nue\ CC interaction sample. The sensitivity to an expected $E_{vis}$ distorsion at low energy has been evaluated to $\sinchooz < 0.07$ at 90\% CL (for $\deltaatm = 2.5 \times 10^{-3} eV^2$ and full mixing). % uncertainties, difficulties \section{Second step} MW-class proton accelerators are being constructed for several physics needs. High intensity conventional horn-focused neutrino beams, ``super beam'', will then provide a new opportunity to further develop neutrino physics: several super beam LBL experiments are proposed as next generation ($\sim$10 years) high sensitivity, high precision experiments before the neutrino-factory era. Their most important goal is discovery of the $\nu_\mu\rightarrow \nu_e$ oscillation but one should not neglect other interests such as detailed study of the neutrino interactions or spin structure of nucleons. In all the future super beam experiments (T2K-I and $\rm{NO\nu A}$), ``off-axis (OA)'' beam plays a key role to achieve high sensitivity: the proton beam line is shooting a few degrees away from the direction to a far detector. In this way, a high intensity low energy narrow band neutrino beam can be obtained and its energy can be adjusted close to the L/E oscillation maximum. Moreover this ``trick'' effectively reduces 2 important background sources: the high energy NC production of $\pi^0$'s and the intrinsic contamination of the beam by $\nu_e$. \subsection{T2K-I} (http://neutrino.kek.jp/jhfnu/) The Tokai-to-Kamioka (T2K) experiment is the next generation LBL experiment in Japan. The $\nu_\mu$ beam is produced using a 50-GeV proton synchrotron from the Japan Proton Accelerator Research Complex (J-PARC). The peak position of its energy spectrum is tunable from 500 MeV to 900 MeV by changing the OA angle from 2 to 3 degrees. The narrow band is important because it increases the fux at the oscillation maximum, maximizing the appearance signal. The far detector is located at Kamioka, 295 km from J-PARC. In the first phase of T2K (T2K-I), the design beam power of the 50-GeV PS is 0.75~MW (more than 100 time as powerful as the K2K beam) and the far detector is Super-Kamiokande (SK) of 22.5-kt fiducial mass. The main purpose of T2K-I is a measurement of $\theta_{13}$ with more than one order of magnitude sensitivity better than any existing experiment ($\simeq < 0.006$ 90\% C.L.). The second goal would be a determination of the ``atmospheric'' parameters, $\theta_{23}$ to an accuracy of 0.01 and $\Delta m_{23}^2$ to $10^{-4}$ $eV^2$. \sinatm\ is presently known to be at least 0.95. Maximal mixing could lead to an underlying new symmetry and thus being able to measure \thetaatm\ to high enough precision to distinguish maximal and nearly maximal mixing is very important. With an OA beam at $2.5^o$ the expected statistics at SK would be 1600 \numu\ CC events per year. The intrinsic beam contamination by \nue\ is 0.4\%. The appearance signal events in SK are interactions with a single showering Cherenkov ring. The neutrino energy is reconstructed assuming quasi-elastic two-body kinematics. An excess of events over the expected background is the signal for \nue\ appearance. Since the signal and background events have different energy spectra, it is essential to control both the flux and the shape of the input spectrum. This is why the experiment will rely on 2 near detectors: at 280m the first near detector will be magnetized (contained in the UA1/NOMAD magnet) and permit detailed studies of the flux and spectra of the different beam components (\numu, \nue, \numubar\ and \nuebar). This knowledge is important in the extrapolation to SK procedure. The second near detector will be situated at 2km, where the energy spectra of the neutrinos crossing a 1kT water Cherenkov and those crossing SK are identical. The uncertainty sources listed above for the K2K experiment should then be drastically reduced. The neutrino beam line construction has started, together with an intensive R\&D and design work on each of its components. The tecnical design report of the 280m detector is expected this summer, while the 2km detector is not yet approved. The first neutrinos in T2K should be delivered in spring 2009 for 5 years. %T2K and NO$\nu$A are complementary experiments. %They use different detector technologies (water cherenkov vs low-Z calorimeter) %but alos their different baseline could help disentangle the effects %of CP violation and matter effects in neutrino oscillations. \subsection{NO$\nu$A} (http://www-nova.fnal.gov/) With the same physics goal as T2K, the NO$\nu$A proposal is in addition raising the neutrino mass ordering question: the mass ordering can be resolved only by matter effects in the earth over long baselines. At 810 km - the NO$\nu$A chosen basline - the matter effect should be about 30\% for a NuMI off-axis beam and only 10\% for T2K. NO$\nu$A is proposed to be a 30 kT totally active low-Z calorimeter (15m x 15m x 130m) placed 15 mrad off the NuMI beam axis. The far detector will consist of 1984 planes of liquid scintillator strips contained in extruded rigid PVC and readout by APD's through wls fibers. Water cherenkov has been discarded because it does not provide sufficient NC rejection at NuMI energy (2 GeV) while a 0.15 $X_0$ sampling calorimeter provides good \pizero-electron discrimination. The fast timing of the scintillator allows to install the detector at ground level: this will mean less than 10 cosmic rays in the 10 $\mu s$ beam spill. The near detector, very similar to the far detector but complemented with a veto and a muon catcher, can fit in several existing locations in the NuMI access tunnel. No single location optimizes all parameters, and the collaboration is considering making it movable or building 2 detectors. NO$\nu$A has been granted stage-I approval by Fermilab in april 2005 and benefits a strong support as being the only approved US experiment in the post 2010 era. The expected beam intensity could reach $\rm 6.5 x 10^{20}$ pot/yr, i.e. 0.65 MW after the collider stops operating in 2009. The construction could start in FY2006, have the first kT operational in 2009 and the full detector operational in 2011. With this optimistic schedule, the experiment is expected to reach an order of magnitude better sensitivity in \thetachooz\ event faster than T2K. %Pending NuSAG/P5 and OMB approval. \section{Third step} T2K and NO$\nu$A will have very limited sensitivity to the CP phase $\delCP$ even if complemented by high sensitivity reactor experiments. A third generation of LBL neutrino experiments will then be required to start a sensitive search for leptonic CP violation. These future experiments will push conventional neutrino beams to their ultimate performances (neutrino SuperBeams), or will require new concepts in the production of neutrino beams. \subsection{VLBL-Brookhaven} The BNL proposal of a Very Long BaseLine conceptual design is advocating the use of a broad band low energy (1-6 GeV) on axis beam heading on a megaton class detector to be sensitive both to \delCP\ and to the sign of \deltaatm. The CP contribution is dependent on both atmospheric and solar $\Delta m^2$ and is affecting the \nue\ appearance spectrum (and \numu\ disappearance) in the 1-3 GeV range. On the other hand, the matter effect causes the \numunue\ conversion probability to rise with energy and is mostly confined to energies $>$ 3 GeV. this energy dependence can be used to measure the value of $\delta_{CP}$ and $\sin^2 2\theta_{13}$. The detector requirements for such an experiment -- both in size and performance -- are well-matched to other important goals in particle physics, such as detection of proton decay and astrophysical neutrinos. In the present design the neutrino beam is produced by the 28~GeV proton beam from AGS (Brookhaven) and is detected by a Mton UNO type water Cherenkov detector in Homestake mine at 2540~km from BNL or in Henderson mine (2700 km). The AGS beam power is supposed to be upgraded to 1~MW from present 0.1~MW by introducing a 1.2~GeV superconducting LINAC for direct injection and increasing repetition rate. The beam is a horn-focused on-axis wide band beam with the spectrum ranging up to about 6~GeV and the peak at around 2~GeV. The expected number of $\nu_\mu$ CC interactions without oscillation is $\sim$13,000/500kt/year. Running with anti-neutrinos could improve further a \delCP\ measurement. \subsection{T2K-II} A future extension of T2K is already envisaged with an upgrading of the proton synchrotron to 4~MW and the construction of a 1-Mt ``Hyper-Kamiokande''. With $\sim$5 times higher intensity and about 25 times larger fiducial mass, statistics at HK will be 2 orders of magnitude higher than at T2K-I. The expected number of $\nu_\mu$ CC interactions is $\sim360,000$/year with a 2$^\circ$ off-axis beam. The goals of T2K-II are the discovery of CP violation and the precise measurement of $\nu_e$ appearance. Preliminary studies on a possible upgrade of the 50-GeV PS to 4~MW have been made by the J-PARC accelerator group. A first gain (by a factor $\sim$2.5) can be obtained by increasing the repetition rate (doubling the number of RF cavities) and by eliminating some idle time in the acceleration cycle. Second, another factor $\sim$2 could be gained by doubling the number of circulating protons when adopting the ``barrier bucket'' method. The design of the neutrino beam line presently under construction at JPARC includes the property of being off-axis (tunable between $2^\circ$ and $3^\circ$) both for T2K-I and T2K-II: the HK site would be in the Tochibora mine, $\sim$8 kilometers away from the SK location at 500 m depth (1400 mwe). Two 250m long parallel tunnels would host huge water Cherenkov detectors similar in principle to SK, amounting to 0.54 Mt fiducial mass. The expected sensitivity on CP violation in T2K-II, based on a full detector simulation (SK scaled to HK), very much depends on the size of the systematic errors. If 2\% error is achieved, then the CP violating phase \delCP\ can be explored down to $\sim 20^\circ$ for $\sin^22\theta_{13}$ greater than 0.01. \subsection{CERN-Fr\'ejus SPL} This new proposal has been stimulated by 2 converging ``opportunities''. First CERN is considering the construction of a new proton driver, a Superconducting Proton Linac of low energy (2-3 GeV/c kinetic energy) but very high intensity (4 MW, i.e. $10^{23}$ protons/yr!). Second the drilling machines of the new safety tunnel in Fr\'ejus should meet at the center around 2009, giving the opportunity to dig a new cavity that could be ready by 2012 and host a Megaton class detector at about 1750m depth i.e 4800 mwe . Although no definite decision on the SPL construction will be taken by CERN before 2009, intense R\&D is already going on for a liquid mercury target station able to cope with the 4 MW beam and for the neutrino beam optics, capable to stand heat, radiation and mercury. Recently an optimization of the SPL neutrino superbeam has been made and found that, given the 130 km baseline (from CERN to the Fr\'ejus tunnel), a 3.5 GeV proton beam plus a 40 m long and 2 m diameter decay tunnel would greatly improve the performances over the 2.2 GeV initial option: the \numu\ CC interaction rate at 130 km would rise from 42 to 122 events/kton/year in case of no oscillation. In these running conditions, both the NC \pizero\ background and the intrinsic \nue\ beam contamination are expected to be low and the sensitivity to \thetachooz\ an order of magnitude better than in T2K-I for a 5 years run of with \numu\ beam (using a 2\% systematic error both on the background normalization and on the signal efficiency). The discovery potential (at 3$\sigma$) to \delCP\ could reach 45$^\circ$ if \sinatm = 0.001 by running 2 years with \numu\ and 8 years with \numubar. Conventional neutrino beams are going to hit their ultimate limitations, specifically in the search for CP violation. But when combined with BetaBeams they can improve the CP sensitivity and allow for T and CPT searches in the appearance mode. \subsection{CERN-Fréjus Beta-beam} The recently proposed beta-beam idea is taking advantage of the possibility of accelerating and storing radioactive ions within their lifetime, thus producing just one flavor neutrino beam (\nue\ or \nuebar). Its energy spectrum is precisely defined by the end point energy of the beta decay and by the $\gamma$ of the parent ion. The flux normalization is given by the number of ions circulating in the storage ring and the beam divergence is determined by the $\gamma$: the beam control is then virtually systematics free. Beta-beam studies are essentially done in Europe presently and synergies with nuclear physics are emphasized. A EURISOL-like complex fed by the SPL could produce $6 \times 10^{18}$ $\rm ^6He$ ions (\nuebar) and $2.5 \times 10^{18}$ $\rm ^{18}Ne$ (\nue) ions per year boosted with a $\rm \gamma = 100$. The superbeam and beta-beam have the advantage of having similar energies which allows usage of the same far detector and explore CP violation in two different channels with different backgrounds and systematics. The disadvantages however are the low cross section at these energies, wich implies very massive detectors, and the limitation in the energy resolution due to Fermi motion. A 10 year experiment, combining a superbeam (running 2 years with \numu\ and 8 years with \numubar) and a beta-beam (running 5 years with \nue\ and 5 years with \nuebar) would give a discovery potential (at 3$\sigma$) to \delCP\ of 30$^\circ$ if \sinatm = 0.001. Ideas about storing radioactive ions that can only decay by electron capture have been recently proposed: this could lead to monochromatic \nue\ beams and should be studied further. \subsection{Neutrino factory} This subject was not discussed in the meeting but could be viewed as the ultimate step for a full understanding of the neutrino mixing and neutrino phenomenology.