161
pose a loss of an interval 12l. It thus appears that the points at which each fringe appears at different distances from the lamina are not placed on a straight line, but on an hyperbola of the second order, which experiment confirms completely.
We must not conclude from this that diffracted light does not move in straight lines, for it is not the same ray that forms a fringe of a given order at different distances. That the ray changes, as the distance is altered, may be concluded from this alone, that the fringes may be observed in space either with the naked eye or with a lens; for then it is evident that the rays which form them must converge, and afterwards diverge; otherwise they could not be collected by the lens so as to afford a visible image of their point of concourse.
Very remarkable phænomena of diffraction are again produced, when the cone of light, instead of being intercepted by an opaque lamina, is transmitted between two bodies terminated by straight parallel edges. In this case, the diffracted fringes may, with great appearance of truth, be attributed to the interference of the two portions of light which fall on the opposite edges.
Nevertheless, there are many physical particulars in the phænomenon, which it is difficult to explain on this hypothesis. M. Fresnel has even found that it is not quite consistent with the measurements of the fringes when they are very exact; he has been convinced that the small portion of light which the edges may reflect is not sufficient to produce the observed intensities of the fringes; and that it is necessary to suppose that other rays assist which do not touch the edges. He has thus been induced to consider all the parts of the direct luminous wave as so many distinct centers of undulations, the effects of which must be extended spherically to all the points of space to which they can be propagated; according to which supposition, the particular effect at each point would result from the interferences of all the partial undulations that arrive at it. This consideration, applied to the free propagation of a spherical wave in a homogeneous medium, makes the loss of light proportional to the square of the distance conformably to observation; but when a part of the light is intercepted, it indicates, in the different points of space towards which it is after-
