[Edi]

I don't know the actual probabilities for common, uncommon and rare sites, but generally common sites appear at least twice as frequently as uncommon ones. It looks like there are about two or three common sites for every uncommon one, and the same seems to hold generally true for the relationship between uncommon and rare. Note that this observation is not reliable, it's more of an educated guess.

I suppose if they have something like 55-60% common, 20-25% uncommon and 15-20% rare, and rare gets always rerolled on the first try, you could get the numbers right that way. Mind you, those were off the top of my head and I have not done statistical analysis (don't know very well how, so my methodology would be off and I suck at statistical math). Kaljamaha is better equipped to answer that question, especially since he did statistical analysis of underwater sites, and the rare sites were really rare.

I can post the results of that later. Doing an overall probability analysis is complicated by the additional factors of terrain restrictions for a lot of sites, which means that frequency of certain types of terrain directly affects the results. Only if you have an equal number of every terrain will the results be reliable.

There's a map (named Sitefinder) Kaljamaha made that he and I used for DB purposes before we got the code from IW, and if anybody wishes to pursue this, we can easily make that available. It is only suitable for testing purposes, but very useful especially for site related testing.

Edi