Spanish grammars written through the 16th and 17th centuries can be used and are to be used as primary sources for the history of the Spanish language. However, when it comes to write the history of the morphology and syntax laid down in them, the following distinctions must be made: the prescribed use or standard, the described uses, and the grammarian’s own written use. In this paper, these distinctions are applied to the history of the analytic future and the conditional tenses (cantar-clitic-he, cantar-clitic-hía), especially in the attempt to explain the loss of these forms in the first half of the 17th century. Nebrija defines the future and conditional tenses as periphrases of infinitive + haber; Correas defines them in the same way, but he does not identify easily the verb haber in the -ía ending of the conditional tense. Neither of them – nor the other grammarians who have been studied: the Anonymous of Lovaina of 1555 and 1559, Villalón, del Corro, Jiménez Patón, Tejeda, Juan de Luna – codify the analytic forms in the verbal paradigms; that means that they do not consider them standard or prescribed usage, although they use them themselves (written uses) and they put them down as examples or other speakers’ uses (described uses). But the written use lasts only until the end of the 17th century; Tejeda and Correas use the analytic forms only in examples (described uses). Taking into account that in the literary language analytic forms last until about 1650, it must be said that the language of the grammars abandons these forms approximately a quarter of a century earlier.These forms were bound to disappear, because of their defective nature and their restricted distribution, and also because they showed a word order (auxiliated + auxiliary), uncompatible with SVO languages. Grammars of the ‘Golden Century’ allow us to understand better their dissapearance right in the first quarter of the 17th century, because they report the full grammatiealization of haber as the only auxiliary verb of the compound tenses.