I don't think that Rg for just a couple residues is meaningful.
Major syntax errors here; you have some reading to do. Within the CORREL module (correl.doc), a few of the time series types use or allow an atom selection (select.doc).
traj firstu 13 nunit 1 select .not. resn tip3 end
The atom selection in the above command within CORREL from your script is meant to select the entire protein, assuming that only water (TIP3) is present. You need to change that and get the syntax and the logic correct, e.g.
traj firstu 13 nunit 1 select segid A .and. ( resid 1 .or. resid 10 ) end
I've assumed the protein is SEGID A; each segment can have a RESID 1, so the RESID alone is not unique for more than a single segment,