The infamous (false) result $$\sum_{n=1}^\infty n=-\frac1{12}$$ is due to Ramanujan. It's obtained by performing illegal manipulations on a divergent sum, basically treating it as if it were a real number. Now, by definition,
$$\zeta(s)=\sum_{n=1}^\infty\frac1{n^s}$$
when $\Re(s)>1$, otherwise we consider the analytic continuation (as the above series would diverge). However, if we were to apply such definition to the case $s=-1$, we would obtain
$$\zeta(-1)=\sum_{n=1}^\infty n$$
and, interestingly enough,
$$\zeta(-1)=-\frac1{12}$$
if we use the correct definition when $s=-1$. This cannot be a coincidence, however I don't see a clear connection between the two results.