The first dictionary prepared and published by the Real Academia Española (RAE) includes citations, mainly drawn from literary works, which show how words are used or even should be used, as the authors they are drawn from are held to be authorities in the use of language (hence the inofficial but widely used name for the dictionary, Diccionario de autoridades). Among the ‘authorities’, there are medieval authors, but most of the works are from the 16th and 17th centuries. Even though the use of the Spanish language by these writers is deemed to be so good as to be held as a model, at the graphic level the text of these works contradict many of the new orthographic rules devised by the Academia. In this paper, I analyze how the academics, contrary to their decision not to alter the traits of the texts used as example (taken after some years or work and the publication of two of the volumes of the six of the dictionary), do in some cases alter the text (ex. in -ava > aba, aver > haver/haber, ç > z) and in others maintain the old graphical traits (ex. in the initial f-, fermosura, fazer > facer...). We study the attitude of the Academia with respect to two different works printed in the 17th century used as ‘authorities’: Alfonso the Wise's Primera Crónica General, read by the academics for their 1604 edition, Gonzalo de Céspedes' El soldado Píndaro, included in its 1696 edition.