Due to the speed of the propagating crack, the indentor is usually still in contact with the flake’s platform before it exits the core.  As the indentor moves obliquely away from the core edge, pulling the flake away with it, a zone of transverse stress can develop across the flake, forcing a bending crack to open along the flake’s length.  The crack initiates at the flake’s dorsal surface and travels towards the ventral surface.  Since the flake is only partly formed at this point, the attached portion ahead of the primary fracture front inhibits the bending at the distal end of the forming flake.  The interaction of the bending force with the propagating crack causes the primary fracture front to divide and propagate in two mostly independent parts.

A ridge or arris starts at the point where this division occurs, resulting in the emergence of a distinctive ‘siret arris’ or ‘split mark’ towards the distal end of the negative flake scar.  Siret fractures are more frequent on flakes struck behind a concavity on the core face because this morphology enhances the bending stress operating on the flake as it forms.  Siret fractures usually occur in hard-hammer percussion flaking but they sometimes occur in soft-hammer percussion and indirect percussion.  They rarely occur in pressure flaking.

Flakes split by siret fractures are generally considered evidence of stone-flaking mistakes by archaeologists, although siret-fractured flakes were used in prehistory as naturally-backed knives; the right-angle siret fracture surface protected the fingers from cuts.  An example was recovered from the Lower Palaeolithic site of Marathousa 1 in Greece, dating to ca. 400,000-500,000 BP.  The split siret surface can also serve as a platform for striking elongated blanks from the edges of the flake, as seen in Australian microlith assemblages.