We double-checked the H-1 NMR of the first fraction, the one that eluted in 40:60 EtOAc/hexanes. This looks like our product in low yield, although it contains two almost equal populations of two similar things, plus impurities. I don't have an explanation for the existence of two closely related molecules, in that both substances seem to have a double bond in the trans-configuration. Another puzzle is why it eluted relatively early. We will continue to think about it.
We could synthesize the dipeptide by deprotecting FMOC at the stage of having a Weinreb amide (not the route that we first tried), which makes a small-scale deprotection marginally more interesting. However, then we face the question of how best to perform the reduction in the presence of the second residue and the peptide bond joining it (a solvable problem, I am sure).