If you are using map
data sets in which area boundaries do not match precisely (for example,
if the boundaries were digitized with a different set of points),
PROC GREDUCE will not be able to identify common boundaries properly,
and this results in abnormalities in your maps. These abnormalities
include mismatched borders, missing vertex points, stray lines, gaps,
and distorted polygons.

If the points in the
area boundaries match up except for precision differences, round each
X and Y value in your map data set accordingly, using the DATA step
function ROUND before using PROC GREDUCE. (See SAS Functions and CALL Routines: Reference for information about the ROUND function.)

For example, if the
map data set APPROX has horizontal and vertical coordinate values
for interior boundaries of unit areas that are exactly equal only
to three decimal places, then this DATA step creates a new map data
set, EXACT, that will be better suited for use with PROC GREDUCE:

data exact; set approx; if x ne . then x=round(x,.001); if y ne . then y=round(y,.001); run;

See About Map Data Sets for additional
information about map data sets.