Now the fact is that in ancient French, and likewise in Provençal, there is still a system of declension more or less independent of prepositions. There are, so to say, three declensions in old French, of which the second is the most important and the most interesting. If we take a Latin word like _annus_, we find in old French two forms in the singular, and two in the plural. We find sing. _an-s_, _an_, plur. _an_, _ans_. If _an_ occurs in the nom. sing. or as the subject, it is always _ans_; if it occur as a gen., dat., or acc., it is always _an_. In the plural, on the contrary, we find in the nom. _an_, and in all the oblique cases _ans_. The origin of this system is clear enough, and it is extraordinary that attempts should have been made to derive it from German or even from Celtic, when the explanation could be found so much nearer home. The nom. sing. has the _s_, because it was there in Latin; the nom. plur. has no _s_, because there was no _s_ there in Latin. The oblique cases in the singular have no _s_, because the accusative in Latin, and likewise the gen., dat., and abl., ended either in vowels, which became mute, or in _m_, which was dropped. The oblique cases in the plural had the _s_, because it was there in the acc. plur., which became the general oblique case, and likewise in the dat. and abl. By means of these fragments of the Latin declension, it was possible to express many things without prepositions which in modern French can no longer be thus expressed. _Le fils Roi_ was clearly the son of the King; _il fil Roi_, the sons of the King. Again we find _li roys_, the King, but _au roy_, to the King. Pierre Sarrasin begins his letter on the crusade of St. Louis by _A seigneur Nicolas Arode, Jehan-s Sarrasin, chambrelen-s le roy de France, salut et bonne amour_. But if we apply the same principle to nouns of the first declension, we shall see at once that they could not have lent themselves to the same contrivance. Words like _corona_ have no _s_ in the nom. sing., nor in any of the oblique cases; it would therefore be in French _corone_ throughout. In the plural indeed there might have been a distinction between the nom. and the acc. The nom. ought to have been without an _s_, and the acc. with an _s_. But with the exception of some doubtful passages, where a nom. plur. is supposed to occur in old French documents without an _s_, we find throughout, both in the nom. and the other cases, the _s_ of the accusative as the sign of the plural. Nearly the same applies to certain words of the third declension. Here we find indeed a distinction between the nom. and the oblique cases of the singular, such as _flor-s_, the flower, with _flor_, of the flower; but the plural is _flor-s_ throughout. This form is chiefly confined to feminine nouns of the third declension.