I understood that a polygon must be defined in a Counter-clockwise direction to be valid. (I got this reversed in my original answer)
A clockwise ring will define a hole in a polygon. But it must be enclosed by the first ring.
So the outside ring must start counter-clockwise with alternate clock/counter-wise nested rings if you had the (real) case of say a lake on an island in a lake.
The check/repair geometry tool will soon take note if you actually manage to build an invalid topology at a low level. Somet loaders check on loading and reject invalid topology. But if you copy a featureclass using featureclassTofeatureclass you will copy the invalid geometry, storing up problems for any later analysis and drawing.
OGC well known text does not define the direction except to note that the righthand rule will imply the polygon 'faces up". If you hod up your right hand with thumb, first and second fingers stretched, first finger along the line, then the second finger points to the inside and the thumb points up.
KML definitely gets it 'wrong' from an Esri point of view, so always repair geometry when loading KML polygons or they won't draw properly and you will have trouble editing them.
I am having trouble with loading data via spatialite too. The WKT loader seems to bypass any geometry validation and even just gets it wrong.
Thanks Kimo.
So, I can assume that all outer rings will be clockwise? I don't care how they are built, by connecting points or converting a dxf, ArcGIS will always have outer rings in a clockwise fashion?
Thanks