proportionate increase of elasticity;
and this accounts for the diminished velocity of light in refracting
bodies. We here reach a point of cardinal importance. In virtue of the
crystalline architecture that we have been considering, the ether in
many crystals possesses different densities, and different
elasticities, in different directions; the consequence is, that in
such crystals light is transmitted with different velocities. And as
refraction depends wholly upon the change of velocity on entering the
refracting medium, being greatest where the change of velocity is
greatest, we have in many crystals two different refractions. By such
crystals a beam of light is divided into two. This effect is called
_double refraction_.

In ordinary water, for example, there is nothing in the grouping of
the molecules to interfere with the perfect homogeneity of the ether;
but, when water crystallizes to ice, the case is different. In a plate
of ice the elasticity of the ether in a direction perpendicular to the
surface of freezing is different from what it is parallel to the
surface of freezing; ice is, therefore, a double refracting substance.
Double refraction is displayed in a particularly impressive manner by
Iceland spar, which is crystallized carbonate of lime. The difference
of ethereal density in two directions in this crystal is very great,
the separation of the beam into the two halves being, therefore,
particularly striking.

I am unwilling to quit this subject before raising it to unmistakable
clearness in your minds. The vibrations of light being transversal,
the elasticity concerned in the propagation of any ray is the
elasticity at right angles to the direction of propagation. In Iceland
spar there is one direction round which the crystalline molecules are
symmetrically built. This direction is called the axis of the crystal.
In consequence of this symmetry the elasticity is the same in all
directions perpendicular to the axis, and hence a ray transmitted
along the axis suffers no double refraction. But the elasticity along
the axis is greater than the elasticity at right angles to it.
Consider, then, a system of waves crossing the crystal in a direction
perpendicular to the axis. Two directions of vibration are open to
such waves: the ether particles can vibrate parallel to the axis or
perpendicular to it. _They do both_, and hence immediately divide
themselves into two systems propagated with different veloci