Suppose you had just made a racemic mixture of #R# and #S# enantiomers, and you wanted to resolve the optical isomers. Because these are chemically identical, there is no easy way or straightforward way. Pasteur (famously) picked out the left-handed and right-handed crystals of lactose with a pair of tweezers, which must have been a very painstaking process. He had no way of knowing that the gross crystalline chirality would translate to molecular geometry, but he lucked out, and here crystalline habit translated into molecular chirality.
So with the #R# and #S# crystals (which might be acidic or basic) you would throw in a homochiral species (a base perhaps), from the so-called chiral pool (the available biological molecules, produced from natural sources, that have a particular chirality - the sugar you put in your coffee is an example). So, suppose I throw in an #S'# stereoisomer, that is avaliable in large quantities. My crystals react with this species, to give a mixture of #RS'# and #SS'# isomers.
#RS'# and #SS'# are no longer enantiomeric. They are diastereomeric (why? because they are NOT mirror images - the mirror image of #RS'# is #SR'#), and are in principle separable by physical and chemical means. The #RS'# and #SS'# should have differential solubility, and in the best circumstances one diastereomer can be crystallized out. This is a very tedious process, but most drugs go thru this process.
An alternative, and very attractive means of separation, would be to perform a so-called chiral resolution, by treatment of the enantiomers with a suitable homochiral reagent. This reagent reacts with ONE stereoisomer preferentially, and the new product may be separated from the chemically different starting material.
