Seeking for one more proof to this claim (namely, that a group of order $255=3.5.17$ is Abelian), I'm looking for an argument (if any) which rules out the case $P\cap Z(G)=\{1\}$, where $P$ is the (one) normal subgroup of order $17$ of $G$. If I could get this, I think that $G/P$ cyclic and $P$ central would lead to the desired result (likewise the usual argument about $G/Z(G)$).
What I know. The nontrivial elements of $P$ must lay in two conjugacy classes (of $G$) of size $3$ and in two other of size $5$ (overall $1+3+3+5+5=17$).
How can I complete this argument?