The poem of Ulrich von Zatzikhoven has scarcely received the attention which, as a factor in the criticism of the legend, it undoubtedly demands. The questions arising out of it are not only interesting, but, as I shall presently show, in one instance at least, of the very highest importance. The questions may be grouped as (_a_) those relating to the structure and sources of the poem itself; (_b_) those which affect its relation to the other Lancelot romances. For the first it is obvious that we are dealing with a poem of very loose construction; the various parts do not harmonise with each other, and no attempt has been made to make them do so. Thus we have no fewer than four love affairs attributed to Lanzelet, and in three out of the four he weds the lady; yet these amours, one of which is subsequent to his marriage with Iblîs, are dropped as of no account. Professor Foerster[23] considers that this looseness of construction points to a late date, and that the source of the _Lanzelet_ was a biographical romance of the weakest order. According to Professor Foerster the clearer the composition, the better knit the incidents, the older the romance. Now it seems to me that there are two orders of ill-constructed romances, and that we shall do well to differentiate between them. In one case we have a number of incidents of secondary character, obviously borrowed or imitated from those occurring elsewhere, strung together more or less cleverly on the thread of a hero's individuality. The incidents are all to be found in other romances, and as a rule none of them have any suggestion of Celtic or mythic origin. The literary style is superior to the matter. Such romances are _e.g._ _Rigomer_, _Torec_, _Le Chevalier à la Manche_. A very favourable example is _Méraugis de Portlesguez_. These are all certainly late romances. In the other case we have a romance even more ill-constructed, but consisting not of incidents but of whole short tales, manifestly independent of each other, and some of them of distinctly antique and mythic character: the literary style is poor and the whole is less a romance, properly speaking, than the material out of which a romance can be evolved. This, I believe, marks an early stage of development, and of this we have naturally but few specimens. The _Lanzelet_ is, I believe, one.