The one you claim you cited.
The first experiment, I believe, was a designed 32 unit protein, specifically chosen because it would act as a template for the designed reaction. It was then split into a 17 unit block and a 15 unit block. The blocks were properly activated and then allowed to react. Voila! A designed experiment produced a designed result. Half a cadillac joined preferentially to the matching other half of a cadillac when superglue was put into the proper position. Fact is, there are plenty of "self-replicating" chemicals. They just reside in living things. Simply put, to dot the i in a penned copy of "Hamlet" does not make you Shakespeare.
We have recently designed one self replicating system based on the leucine zipper motif of GCN4. One 32-residue alpha-helical peptide serves as a template to organize two constituent fragments in the proper orientation prior to ligation. Condensation of the two fragments produces a second template which can serve as a template for another such reaction