We consider entanglement-swapping schemes with general (rather than maximal
ly) entangled bipartite states of arbitary dimension shared pairwise betwee
n three or more parties in a chain. The intermediate parties perform genera
lized Bell measurements with the result that the two end parties end up sha
ring an entangled state which can be converted into maximally entangled sta
tes. We obtain an expression for the average amount of maximal entanglement
concentrated in such a scheme and show that in a certain reasonably broad
class of cases this scheme is provably optimal and that, in these cases, th
e amount of entanglement concentrated between the two ends is equal to that
which could be concentrated from the weakest link in the chain.