As to colour, we must refer as much as possible to oil-painting. In fact, the colour of the human skin, as we have formerly said,[354] is, in reality, a complex visual impression; all the coloured rays (we employ the term here in the conventional sense given to it in physics) which emanate from the skin, and which strike the eye of the observer, are not formed by the same plane surface; they arise from the more or less profound parts seen by transparency, through a more or less diaphanous medium, more or less favourable for the emission of these rays. Hence results, as regards the eye, a special sensation, and as regards the mind, a special notion, which we explain in the arts by the word _transparency_ or _diaphaneity_. Now, this kind of sensation will not be reproduced by the artist unless he employs certain processes recalling to the mind those of nature itself. This is not the case with water-colour painting. The colouring matter, reduced to extremely fine particles, is applied, it is true, in a transparent vehicle--water; but this, destined to evaporate almost immediately, leaves the colour on the surface of the paper, stretched into an extremely fine layer, without appreciable thickness. We perceive from this the radical imperfection of water-colour for portraiture, and the impossibility of rendering by such means, at least with truth, the effect of skin colours. Oil painting offers far better resources, and here is the secret of its incomparable superiority. The colouring matter, diluted by the oil, remains suspended as before in its transparent medium when the painting is dry; so that the luminous rays, in order to arrive at the eye, start from the surface of the paint as well as from its interior substance. We find exactly the same process in nature; an impalpable powder, like the pigmentary granulations, or the globules of blood in the capillaries of the skin, is spread over a diaphanous substance. We may now understand the advantage of such a process in anthropological iconography. We must, indeed, almost give up all other methods. It is easy to convince oneself of the fact by examining the coloured portraits which illustrate the works of Prichard,[355] Nott, and Gliddon,[356] who are, however, extremely particular about the correctness of the types which they bring before our notice. But all these coloured portraits are unsatisfactory, and when we see some anthropologist invoke the authority of these bad prints, we really ask ourselves which we ought most to admire, either the blind confidence of the _savant_, the imprudence of the author, or the rashness of the artist. Fancy, however, attacking with such platitudes the portraits of dark-coloured men which the masters of painting have left us, from Veronese to Géricault! They alone have been able, by their process, to seize the reality of the complexion and colour of their models.[357]