We use space motions of globular clusters as a new constraint to deter
mine the mass distribution of our Galaxy. Among the 26 globular cluste
rs with known radial velocity and proper motions (Dauphole et al. 1994
), eight of them are now observed at distances smaller than or nearly
20 kpc from the Galactic centre with a velocity that throws them at a
distance greater than 40 kpc, with the usual galactic potentials. More
over, the time spent at large distances from the Galactic centre corre
sponds to a large part of the orbital period. Statistically, we must n
ormally find a lot of globular clusters at these distances. But the di
stribution of the globular clusters shows a gap beyond 40 kpc. We use
this contradiction between observations and theory as a new constraint
to build a potential model of our Galaxy, that is globular clusters w
ith known space velocities must not have apocentric distances greater
than 40 kpc or in other words the potential must be strong enough to p
revent galactic globular clusters to go beyond this limit. The derived
model is simple, completely analytical and works quite nicely with th
is large scale constraint. Total mass of the model is 7.9 10(11) M(cir
cle dot). A large part of this mass must be distributed between 15 and
40 kpc and must be in fact dark matter.