rearranged. These four samples, although of the same period, are of the two different types already referred to. The outer ones are of the transition style, in which the face has passed quite out of the hideous, into the laughing, mocking kind; while at the same time it preserves traces of the archaic in the protruded tongue with indentation in the middle, indicating the split. The two faces between are fair examples of the Roman Medusa, and of that commonly seen as the product of the most modern foundries of to-day. All these heads are of course of approximately the same date, and show that at the time of Pompeii it was the fashion to give them wings; but it will be noticed that only the latest type shows here any sign of the conventional snakes, and I may state at once that very rarely have I found snakes upon heads retaining any remains of the grinning mouth or protruded tongue. Upon the Parthenon there is a Gorgon's head of somewhat archaic type, with snakes beneath the chin, not on the head; this remarkable exception indicates, as I think, that these snakes were an addition not found on older examples; and shows probably the very first, at any rate the earliest known, representation of what was even then a late development. It may be remarked that the representation of the hair developed into snakes can hardly be found in any early representation, say—before the time of Praxiteles. Moreover, on this exceptional example the snakes are not a development of the hair proper, but are beneath the chin. On this point we shall have more to say.
As I propose to work back from late to early examples, I have given the Pompeian heads as fair specimens of the Medusa of to-day, and I further draw attention to the modern arms of Sicily (Fig. 1, Plate VI.,) which show both wings and snakes.
One of the most exaggerated and fantastic forms of the snake development is that known as the Gnostic
