Ron brings up a very good point. Proper lab technique is crucial but, is often times lacking. I have found that the ASTM method works well for sands and gravels but becomes difficult to vague with clayey silts and near impossible with more plastic clays.
As I mentioned above, the addition of significant amounts of sulfate (gypsum) makes for some very interesting answers. Volcanics and some ores can throw things way off.
For most commen (non-specification) applications I use the Shrinkage Limit calculations for plastic soils. I have also used several methods of back calculating to 'guess' the Specific Gravity I do not believe that most computations require a Specific Gravity closer than 0.03 (resulting in a sequence of 2.60, 2.63, 2.66 ....) but enough check tests should be made to be reasonably sure that strange soils or materials are not being encountered and ignored.