For a decade, technological or natural networks have appeared to have a common mathematical architecture. This type of architecture has a node connectivity which follows a power law distribution. This architecture confers to these networks a resistance property to the loss of nodes. Such properties are advantageous for evolutional networks through time. Thus, this architecture can be expected in animal social networks. Another characteristic commonly met concerns the structuration of the network into communities by the mechanism of assortative mixing by vertex degree (i.e. by the number of ties individuals have). Such a structure is a reflection of evolutional mechanisms: the preferential attachment and the triadic closure processes. Using recent analytical techniques on an affiliative social network in a non-human primate species (Macaca sylvanus), we analysed the mathematical architecture and its properties. We demonstrate that in spite of the use of a recent protocol supposed to permit this type of analysis, the type of distribution cannot be clearly determined, encouraging us to carefully interpret the results obtained until then. Nevertheless, we observed interesting properties of the network at an ecological and evolutional level with network resilience that allows a cohesive society to be maintained even when faced with a catastrophe (high predation, epidemic).