Take the symmetrical form of the completed Zeta-function:
$\displaystyle \chi(s) \zeta(s) = \chi(1-s) \zeta(1-s)$
with
$\chi(s)=\pi^{-(\frac{s}{2})} \Gamma(\frac{s}{2})$.
For $s=\sigma + ti$, I conjecture that only for $\sigma=\frac12$:
$\Re(\chi(s)) = \Re(\zeta(s)) =0$ is always true, except when $s=\rho_n$ (assuming RH).
If $\sigma=\frac12$ and $t$ is real then $\Re(\chi(s))=0$ can be rewritten as:
$A(t) = \cot \left( \frac12 t\ln \left( \pi \right) \right) \dfrac{\Re \left( \Gamma \left(1/4+\frac{t i}{2} \right) \right)}{\Im \left(\Gamma \left( 1/4+\frac{t i}{2} \right) \right)} +1 = 0$
Similar to $s=\rho$ for non trivial zeros, let's call $s=\alpha$ when $A(t)=0$.
There seems to be a strong connection between the $\Re(\zeta(\alpha_m))$ and $\Re(\zeta(\rho_n))$. They exclusively come in an adjacent pair $(\alpha_m,\rho_n)$ or $(\rho_n,\alpha_m)$ that is connected via a 'sharp trough' of $\Re(\zeta(s))$ through the x-axis. The distance between the paired values appears to become smaller when $t$ grows.
If this paired pattern is true, then it would imply that when $\Re(\zeta(s))$ dives into a 'trough' through the x-axis and we find that $s =\alpha_m$, then one could predict with certainty that $s=\rho$ for the next time $\Re(\zeta(s)) = 0$ and that it must be located before $s=\alpha_{m+1}$. In that sense there would be information hidden in $\chi(s)$ that constrains the location of the $\rho$'s.
A bit complicated question, I know, but hope the picture below illustrates the thought.
http://imageshack.us/photo/my-images/855/riemannzero.jpg/
Questions:
Has it been proven that $\Re(\chi(s)) = \Re(\zeta(s)) =0$ (and/or $\Im(\chi(s)) = \Im(\zeta(s)) =0$) only when $\sigma=\frac12$ and $s \ne \rho$ ?
Is there anything known about $\alpha_m$ and $\rho_n$ always coming in connected pairs $(\alpha_m,\rho_n)$ or $(\rho_n,\alpha_m$) and/or that they converge when $t \rightarrow \infty$?
Thanks.
One additional afterthought:
Assuming RH and the adjacent paired values of $Re(\zeta(s))=0$ always having the shape:
$(\alpha,\rho)$ or $(\rho,\alpha)$,
then it would only require counting $\alpha$'s (from $t>1$) to establish the exact number of $\rho$'s $\pm1$ below a certain number $N$ (i.e. without even looking at $\zeta(s)$).