We describe an experimental procedure to mimic the formation of long (over 40 residues) co-oligopetide sequences in many identical copies which may have occurred in the prebiotic molecular evolution. The basic hypothesis is that chain formation is based on the stepwise fragment condensation of randomly generated short oligopeptides, whereby the elongation takes place under the contingent environmental constraints (solubility, pH, salinity), which eliminate most of the products, and thus determine the selection towards one particular small set of chains. The present work aims at verifying the validity of this scheme. In order to do so, we utilize a classic synthetic procedure based on the Merrifield solid-phase synthesis of peptides for the synthesis of randomly produced peptides as well as for their stepwise fragment condensation. Thus, starting from a library of peptides with n=10, the first condensation step produces a library of 16 peptides with 20 residues each (n=20), of which only four remain water-soluble and, therefore, capable to undergo the next fragment condensation step. This gives rise to 16 peptides with n=30, out of which twelve precipitate out under the chosen pH and buffer conditions and are eliminated. Finally, a 44-residue-long water-soluble de novo protein is obtained. This has no homologies or similarities with extant proteins, and, based on circular dichroism (CD), it assumes a stable three-dimensional folding. In agreement with CD data, molecular-modelling simulations suggest an helical fold for the protein with poor, if any, structural homology with known proteins. The implication of this procedure as a general mechanism for the etiology of de novo macromolecular sequences and globular proteins in the origin of life is briefly discussed.