In this work, we present new data and an original hypothesis regarding Martin Harriet�s grammar (1741), and in particular concerning the two vocabularies �Basque-French (BF) and French-Basque� that it contains. First, we will offer the testimony of the author himself regarding his work, larger and more ambitious than usually acknowledged (1.1). Furthermore, we will place the vocabularies in the context of the history of Basque lexicography, by comparing the size of his nomenclature with other dictionaries of the time (1.2). Then, we will show that a formal analysis reveals that BF comes from a French Basque dictionary similar to FB, but probably larger than that one. This allows us to say that FB is not a mere re-elaboration of BF, the hypothesis currently admitted. We speculate that both dictionaries are the outcome of a separate selection of lexical entries having their origin in the hypothetical bilingual dictionary proposed (2 and 3). Finally, we show that Harriet used as a source some previous works: besides Axular, that we already knew, we also find traces of Voltoire, Leizarraga and Xurio at least (4).