The most reliable method to estimate the residence time of cosmic rays in the Galaxy is based on the study of the suppression , due to decay , of the flux of unstable nuclei such as beryllium–10 , that have lifetime of appropriate duration .
The Cosmic Ray Isotope Spectrometer ( CRIS ) collaboration has measured the ratio between the fluxes of beryllium–10 and beryllium–9 in the energy range E _ { 0 } \simeq 70 –145 MeV/nucleon , and has used the data to estimate an escape time \tau _ { esc } = 15.0 \pm 1.6 Myr .
This widely quoted result has been obtained in the framework of a simple leaky–box model where the distributions of escape time and age for stable particles in the Galaxy are identical and have exponential form .
In general , the escape time and age distributions do not coincide , they are not unique ( because they depend on the injection or observation point ) , and do not have a simple exponential shape .
It is therefore necessary to discuss the measurement of the beryllium ratio in a framework that is more general and more realistic than the leaky–box model .

In this work we compute the escape time and age distributions of cosmic rays in the Galaxy in a model based on diffusion that is much more realistic than the simple leaky–box , but that remains sufficiently simple to have exact analytic solutions .
Using the age distributions of the model to interpret the measurements of the beryllium–10 suppression , one obtains a cosmic ray residence time that is significantly longer ( a factor 2 to 4 depending on the extension of the cosmic ray halo ) than the leaky–box estimate .
This revised residence time implies a proportional reduction of the power needed to generate the galactic cosmic rays .