This paper presents experimental results on the fragmentation of a low meting point liquid metallic alloy jet into water. The liquid is Field's metal whose melting point is 62 • C. Data are obtained using high-speed camera acquisition and the solidified particles are sieved, a size Probability Distribution Function (PDF) is obtained from their mass distribution. These results are compared to separate data acquisitions obtained using a phase Doppler anemometer (PDA used in reflexion regime). Injection diameter range from 1 mm to 5 mm and injection velocity from 2.28 m/s to 4.97 m/s resulting in a (carrier phase) Weber number ranging from 15 to 281 and a (carrier) Reynolds number ranging from 2500 to 24000. The conclusion is that for these intermediate Weber and Reynolds numbers, the size of the droplets can mainly be related to a cascading instability mechanism: first Entov & Yarin bending instability which is then closely followed by a Rayleigh-Taylor instability. Moreover, the Mass Probability Distribution Function (PDF) can be approximated by a log-stable laws whose parameters can be computed using wavelengths stemming from the cascade of instability scenario. However, the smallest droplets or ligaments can eventually reach the turbulent dissipation length-scales where they are re-agglomerated by the turbulent movements. This results now in a Number PDF that also follows a log-stable law. For this part of the distribution, it is also possible to compute the parameters of the distribution without using any fitting parameter, by estimating the different turbulent scales. In this setting, atomization looks like a competition between a direct "cascade" of fragmentation instability and an inverse "cascade" of turbulent re-agglomeration PACS numbers: May be entered using the \pacs{#1} command.