P. de B. St. proposes a new dialectology for Ancient Celtic languages, by stressing contacts and common innovations. She also proposes an earlier date for the introduction of Celtic in Spain and Italy : the presence of Celtic (non Gaulish) elements in the Iberian corpus, and in Ligurian would lead to suppose a multiplicity of Celtic layers in both peninsulas. At least, five successive layers of Celtic languages could then be distinguished. Exactly as there were Celts in Spain before the Celtiberians, in the same way in Northern Italy, Lepontic was preceded by a Celtic element present in Ligurian : Ligurian itself would be a Celtic language, partly preserving of Indo-European -p-, and presenting some other remarkable archaisms. The author is...