We study the clustering of galaxies in real and redshift space using the Optical Redshift Survey ( ORS ) .
We estimate the two point correlation function in redshift space , \xi ( s ) , for several subsamples of ORS , spanning nearly a factor of 30 in volume .
We detect significant variations in \xi ( s ) among the subsamples covering small volumes .
For volumes \buildrel > \over { \sim } ( 75 h ^ { -1 } { Mpc } ) ^ { 3 } , however , the ORS subsamples present very similar clustering patterns .
Fits of the form \xi ( s ) = ( \frac { s } { s _ { 0 } } ) ^ { - \gamma _ { s } } give best-fit values in the range 1.5 \leq \gamma _ { s } \leq 1.7 and 6.5 \leq s _ { 0 } \leq 8.8 h ^ { -1 } Mpc for several samples extending to redshifts of 8000 km s ^ { -1 } .
However , in several cases \xi ( s ) is not well described by a single power-law , rendering the best-fit values quite sensitive to the interval in s adopted .
We find significant differences in clustering between the diameter-limited and magnitude-limited ORS samples within a radius of 4000 km s ^ { -1 } centered on the Local Group ; \xi ( s ) is larger for the magnitude-limited sample than for diameter-limited one .
We interpret this as an indirect result of the morphological segregation coupled with differences in morphological mix .
We split ORS into different morphological subsamples and confirm the existence of morphological segregation of galaxies out to scales of s \sim 10 h ^ { -1 } Mpc .
Our results indicate that the relative bias factor between early type galaxies and late-types may be weakly dependent on scale .
If real , this would suggest non-linear biasing .

We also compute correlations as a function of radial and projected separations , \xi ( r _ { p } , \pi ) , from which we derive the real space correlation function , \xi ( r ) .
We obtain values 4.9 \leq r _ { 0 } \leq 7.3 h ^ { -1 } Mpc and 1.5 \leq \gamma _ { r } \leq 1.7 for various ORS samples .
As before , these values depend strongly on the range in r adopted for the fit .
The results obtained in real space confirm those found using \xi ( s ) , i.e .
in small volumes , magnitude limited samples show larger clustering than do diameter limited ones .
There is no difference when large volumes are considered .
Our results prove to be robust to adoption of different estimators of \xi ( s ) and to alternative methods to compensate for sampling selection effects .