In two great caves decorated in the Paleolithic : Montespan and Le Tuc d’Audoubert (out of which a river flows) there were found headless snake skeletons. One of them was a Colubrid, a particularly long snake. The fact that the skeletons are headless may reflect the snake’s dangerousness : the dangerous animals, like bison and lions, were often represented arrowed or headless, for instance at the Tuc d’Audoubert cave. To understand the meaning of what was possibly an important ritual I have tried to statistically reconstruct the primal European Paleolithic folklore about the Snake. The software Paup4.0a147 and Mesquite2.75 were used to analyze a database of 42 mythological narratives concerning snakes (available here : http://ruthenia.ru/folklore/berezkin) and 22 geographical areas, each of them possessing more than ten narratives. The use of such a database was justified by a previous phylogenetic study of folktales concerning wild animals (d’Huy 2015) that showed a low correlation between the distribution of folk tales and geographical associations among populations, and established a parallel between the structure of phylogenetic trees built from folk tales and what we know of the first human migrations ; moreover, these phylogenetic trees allowed the reconstruction of the content of ancient tales corpora, e.g. Paleolithic. In this paper, a NeigbhorJoining (retention index : 0,52; fig.1), an UPGMA (retention index : 0,48; fig.2) and a majority rules consensus (RI : 0.54; fig.3) trees were build. The organization of these trees are similar to those obtained by the analysis of three other different databases : the first is about the stories where a rainbow snake drinks water on the ground, or in the sky (d’Huy on 2016 ; fig.4) ; the second, the fight against the dragon, based on many ethnic group’s beliefs (d’Huy 2014a ; fig.5) ; the third is based on some geographical areas and concerns the zoem of the dragon (d’Huy, 2013 ; fig.6). The results from these four independent, unrelated databases mostly coincide with respect to the topology of branching, and converge to strong conclusions about the evolution of snake folklore. Four almost identical trees have been reconstructed from the databases of the present paper (fig. 1, 2 and 3) and of d’Huy 2013 (fig.6). The human folklore about snakes seems to have left Africa in two steps. The first wave reached Oceania and then South America through the South and Southeast coast of Asia. This reconstructed wave is in agreement with the diffusion of the particular motif where a snake rainbow drinks water on the ground, or in the sky (fig. 4). Another wave left Africa to the Americas through central Asia (?). Finally, a late wave covered the major part of Eurasia, perhaps soon after the late glacial maximum. These two last waves are reflected in the two-step diffusion of the fight against the dragon (fig.5). The results obtained from four different databases at distinct levels (the diffusion of a set of narratives, a set of traits about a zoem, a mythological motif, a narrative structure where different story elements serve the same narrative plot) using different levels of analysis (geographical area, cultural area, cultural groups) and different statistical tools (including NeigbhorJoining, UPGMA, Bayesian algorithms, parcimony, consensus) all fit the same scenario. Once the structural integrity of the NeigbhorJoining, the UPGMA and the consensus trees controlled by the three independent sources of evidence, it is important to note the strong closeness between the structure of these mythological trees and what we know about the first human migrations. It seems possible to accurately reconstruct the first human migrations from the diffusion of sets of folk tales, or of given mythological narratives, by using statistical tools. Using all phylogenetic trees studied in this paper and monitoring the results by observing the congruence between the results and the reconstructions obtained from two completely different databases (d’Huy 2013, 2014a, b) at the same geographical level, it becomes possible to securely reconstruct various features of the first European folklore about snakes. During the European Paleolithic period, people imagined mythological snakes being of natural size (100 % probability), or giant (75 % probability), possessing horns on the head (83 % probability) and having one or many heads (100 % probability) ; they sometimes possessed a diamond or magic treasure, could fly and produce rain and/or thunderstorms (75 % probability) and could fight against the thunder or against a giant bird (66.66 % probability). They could take the form of a rainbow (75 % probability). They were dangerous. They could prevent people from reaching water : permitting them, most of the time, only in exchange for sacrifices or valuable goods (100 % probability). Snakes were not only immortal, but were responsible death, for the mortality of man (100 % probability). And so it is possible that the headless snake skeletons of Montespan and of Le Tuc d’Audoubert reflect a ritual used to control rain and water by neutralizing, symbolically, the dangerous master : the dispenser of this precious element. This reconstructed myth may well have been illustrated by a Paleolithic image found in the cave of La Madeleine, in Dordogne (France), dated to the Magdalenian period. This image (fig.7, 8), possibly related to water, shows an apparently dead snake thrown on his back, passive, on which a man with a stick on his shoulder turns his back. Significantly, the heads of two horses are added. Several current mythological motifs (shown to be, in fact, very old) and the comparison with other Paleolithic images (e.g. fig. 9) may give us an explanation. The horses as auxiliaries in the fight against the monster and the passivity of the defeated snake are motifs that are found among the first Indo-European narratives. Moreover, water is often associated with snakes in Paleolithic images (and are among, once again, the first Indo-European narratives). This second part of the paper is meant as independent support, and does not use statistics.