Context : All Globular Clusters ( GCs ) studied in detail so far host two or more populations of stars ( the multiple population phenomenon ) .
Theoretical models suggest that the second population is formed from gas polluted by processed material produced by massive stars of the first generation .
However the nature of the polluter is a matter of strong debate .
Several candidates have been proposed : massive main-sequence stars ( fast rotating or binaries ) , intermediate-mass AGB stars , or SNeII .

Aims : We studied red giant branch ( RGB ) stars in the GC M4 ( NGC 6121 ) to measure their chemical signature .
Our goal is to measure abundances for many key elements ( from Li to Eu ) in order to give constraints about the polluters responsible for the multiple populations .

Methods : We observed 23 RGB stars below the RGB-bump using the GIRAFFE @ VLT2 spectroscopic facility .
Spectra cover a wide range and allowed us to measure light ( Li , C , ^ { 12 } C/ ^ { 13 } C , N , O , Na , Mg , Al ) , \alpha ( Si , Ca , Ti , ) , iron-peak ( Cr , Fe , Ni ) , light-s ( Y ) , heavy-s ( Ba ) , and r ( Eu ) elements .
We completed this study by analyzing a subsample of the UVES spectra presented in Marino et al .
( 2008 ) in order to have further clues about light s-elements of different atomic number ( Y and Zr ) .

Results : We confirm the presence of a bimodal population , first discovered by Marino et al .
( 2008 ) .
Stars can be easily separated according to their N content .
The two groups have different C , ^ { 12 } C/ ^ { 13 } C , N , O , Na content , but share the same Li , C+N+O , Mg , Al , Si , Ca , Ti , Cr , Fe , Ni , Zr , Ba and Eu abundance .
Quite surprisingly the two groups differ also in their Y abundance .
This result is strongly supported also by the analysis of the UVES spectra .

Conclusions : The absence of a spread in \alpha -elements , Eu and Ba makes SNeII and AGB stars unlikely as polluters .
On the other hand , massive main-sequence stars can explain the bimodality of Y through the weak s-process .
This stement is confirmed independently also by literature data on Rb and Pb .
The lack of a Mg/Al spread and the extension of the [ O/Na ] distribution suggest that the mass of the polluters is between 20 and 30 M _ { \odot } .
This implies a formation time scale for the cluster of 10 \div 30 Myrs .
This result is valid for M4 .
Other clusters like NGC 1851 , M22 , or \omega Cen have different chemical signatures and may require other kinds of polluter .