The leaving out of sight of all methods and materials other than those employed for the immediate purpose in hand has resulted in most unfortunate, and in some cases ridiculous, crests which could never have been used in the ancient way, and now if they have to be carved in relief or in the round, as mural decoration or as the newel of a stairway, show themselves wholly inadaptable to reasonable treatment. On the other hand the early crests are always “possible,” for the mediaeval herald was naturally familiar with the appearance of an actual crest modelled in the round, though he may never have modelled one himself, and so his design is always structurally right. But what can be said for some modern examples, a dove flying over water for instance? It seems to have been forgotten until recently that heraldry ever had a real existence or could possibly be carried out in more than one way, and the result was that anything that was suitable to a flat shield was thought equally appropriate for a crest so long as it was sufficiently differenced from other bearings. A few experiments with a lump of clay would have shown the fallacy of this idea, and incidentally might have saved many a family, often in spite of itself, from being labelled for ever with an absurd bearing. As, however, we cannot always choose the heraldic motives with which we have to deal we must make the best of the refractory ones, as well as of the rest, and the structural side of the subject may be regarded as the direction in which the solution of difficult problems may be found. As an example, let us take a rampant lion and pose it on a helm, and it becomes obvious that if it is taken from a shield without modification it will look ridiculously insecure on one leg as it is generally posed, Fig. 150, but that it is much improved if arranged in firmer relation to its base, the helm, as in Fig. 151. [Illustration: FIG. 150.] [Illustration: FIG. 151.] Another difficult kind of crest is that which is differenced with a row of objects in front. For instance: In front of a tower between two wings three fleurs-de-lis; which is usually drawn as Fig. 152, although wings would never have been modelled fore and aft in that way. If the solidity of a crest is kept in mind the model would come out something like Fig. 153, and on similar lines the most unpromising material may with a little pains be made presentable by the use of sketch models in the round. [Illustration: FIG. 152.] [Illustration: FIG. 153.] When two helmets and crests occur together they turn towards each other and so naturally help the unity of design, and when there are three, the outer ones turn similarly towards that in the centre. If, however, the crest be of such a nature that it cannot be turned about, it will of course be preferable for the crest to dictate the position of the helm rather than to repeat the senseless fault of the helm and crest facing different ways.