1919] OF SOME BRILLIANT ANIMAL COLOURS 591 certainly there is no improbability in surface-reflexion playing a part. It may be that both causes are operative in a single specimen and even at the same part of it. The next contribution to the discussion is an important one by Mallock*, who brings to bear the instinct and experience of a naturalist as well as of a physicist. His observations were mainly on the feathers of birds and the scales of insects, and they lead him to regard interference rather than selective reflexion as the origin of the iridescent colours. " The transparency or, at any rate, the vanishing of the characteristic transmitted colour in the case of all animal tissues when immersed and permeated by a fluid of the same refractive index is strongly in favour of interference being the source of the colour, but even stronger evidence is given by the behaviour of the structures under mechanical pressure. " If the grain or peculiarities which favour the reflexion or transmission of particular colours is of molecular size, there is no reason to suppose that pressure insufficient to cause molecular disruption would alter the action of the material on light. On the other hand, if the colours are due to interference, that is, to cavities or strata of different optical properties, compression would alter the spacing of these, and thus give rise either to different colours or, with more than a very slight compression, to the transmission and reflexion of white light. " In every experiment of this kind which I have made either on feathers or insect scales the effect of pressure has been to destroy the colour altogether.... " With many feathers the colour returns when the pressure is taken off, but with insect scales the structure seems to be permanently injured by compression, and though when allowed to expand again the material is not colourless the brilliancy which belonged to the uninjured scale is gone, and the colour in general changed. "The facts above mentioned seem to offer stronger reasons in favour of interference than the polarization phenomena referred to by Michelson and Walter do against it." I have already commented on the importance of the evidence afforded by observations with polarized light; and if we have to choose between selective reflexion and thin plates of the type usually considered in theoretical writings, we may find ourselves in a position of much difficulty. The question then arises, Is there any loophole for escape ? I think there may be. The polarizing angle, as given by Brewsber's law, depends much upon what we may call.the smoothness of the reflecting surface. A moderate curvature is of no significance in this connexion, but when the-radius of curvature- becomes co'mpar'able with the wave-length of the light it is another matter. Thus in the case of smooth glass the polarizing angle is about 5*7°that"is, light * Proc. Roy. Soc. A, Vol. LXXXV. p. 598 (1911). - ' grams are rather confusing in that they suggest a phase-difference of 180° between'the two polarized components reflected perpendicularly, when evidently the distinction between the two components disappears. -.;-6,1807). If the mercury be wet, boiling may be dispensed with and negative pressures of two atmospheres are easily demonstrated.