For those getting tired of the emunos v'deyos topics (namely, me), back to the world of gemara. Reuvain sells Shimon an ox. After purchase, Shimon discovers that the ox will not plow his field, so he comes back to Reuvain and demands a refund for his lemon. Most people buy oxen to plow, and this ox is clearly not in the mood to work. Reuvain responds that the ox is certainly usable -- it can be shechted and made into a nice steak dinner. It's not his fault that Shimon (like most people) insists on using his ox only to plow, and he (Reuvain) never agreed to those terms in the contract. Rav holds in this case a refund is owed because we determine the meaning of the contract based on rov and rov people buy oxen to plow. Shmuel, however, disagrees and holds "ain holchin b'mamon achar harov" (B"K 46, B"B 92).
Rashbam (Baba Basra 93a d"h d'hu gufei) writes that Shmuel's principle only holds true when we speak of a ruba d'leisa kaman, a rov based on behavior of the masses.
Tosfos in a number of places implies that Shmuel's principle applies even to a ruba d'isa kaman, a rov based on the statistics before us. For example, the gemara (Kesubos 15) writes that a foundling in a city where the majority of the residents are Jews is treated as a Jewish child even with respect to monetary halachos like returning a lost object to him. Tosfos questions how this gemara fits Shmuels dictum that "ain holchim b'mamon achar harov". In this case we are not dealing with a behavioral phenomenon like why people buy oxen but rather we are dealing with the statsitical liklihood of a child being Jewish based on an accounting of the residents of a specific city (see Ch. R' Akiva Eiger Sanhedrin 3b for another Tosfos along these lines).
According to the Rashbam, there is no question from this gemara on Shmuel. Shmuel rejected only ruba d'leisa kaman as proof, but not ruba d'isa kaman.
My son and I were discussing last night what the point of disagreement between Tosfos and the Rashbam might be. Floor is open for comments before my 2 cents get thrown in.