Bone breakage for marrow extraction is a common strategy in human groups and societies belonging to different chronological cultures. To understand long bone breakage, especially in Palaeolithic contexts, many studies use experimental archaeology as a mean to document these subsistence strategies. We have initiated new experiments on this topic, both to consider the morphological variability of marks and fractures, but also to document the interface between bone material and hammerstone types. The purpose of our study was to create a referential of fractured bovid long bones (humerus, radius-ulna, femur, tibia; n=80) according to a controlled protocol (Viallet et al., this conference) and to analyze the first sample of this corpus. The quantification of bone remains distinguishes large-size to small-size fragments (potentially non-determinable) on the 22 bones studied to evaluate the number of remains (NR) produced according to the element and percussion technique, their size distribution, but also to qualify the fractures types (spiral, longitudinal, oblique, transverse) and other characteristics (e.g. inner adhering flake). A precise morphological description and variations of the impact marks (completed: notch, or not) is given to demonstrate a potential correlation between the number of percussion marks, NR, hammerstone types, and bone structure. The first results show that the femur is the easiest bone to break (elongated and cylindrical diaphysis) with an optimization of the number of blows, and the production of a limited number of splinters of large dimensions. Conversely, the structure of the radio-ulna seems to influence its fracturing, involving the greatest number of percussion marks and production of more small-sized remains. Between these two extremes, the tibia and humerus show a differentiated spectrum (NR and size, percussion marks). Detailed plates, accompanied by photographs and a precise analysis are proposed in our contribution.