TRANSACTIONS OP SECTION A. 0*95 



of long copper wires spanned 6 centimetres apart. A short length of these wires 

 was then immersed in an electrolyte, and the intensities of the transmitted wave 

 trains for several different thicknesses of the electrolyte layer compared by an 

 electrometer. If the altered intensity of the wave train were due solely to absorp- 

 tion in the conducting layer, it would be easy from such data to calculate the 

 conductivity of the solution. 



The matter proved, however, to be not so simple. The transmitted intensity 

 did not regularly decrease, but varied periodically. The effect was obviously 

 analogous to, or rather identical with, the interference phenomena of thin plates. 

 The transmitted ray is a minimum for a plate a quarter wave-length thick, a 

 maximum for a plate a half wave-length thick, &c. 



Owing to this, and a further experimental complication arising from the 

 multiple reflection of waves between every pair of reflecting points on the circuit, 

 the method became too complex to permit of calculating the conductivity of the 

 solution, but the phenomenon retained sufficient intrinsic interest to warrant more 

 complete investigation. Curves showing the transmitted intensity as the thickness 

 of the layer was increased were determined for water, dilute solutions of zinc 

 sulphate, 95 per cent, alcohol, and a mixture of alcohol and water. In all cases 

 the first maximum is very well marked. As this first maximum gives us the half 

 wave-length in the liquid, the method can be used for determining dielectric 

 constants. The following table gives those thus determined for the liquids men- 

 tioned above: — 



Distilled water .... 69-5 | 95 per cent, alcohol . . , 26-7 



(1) ZnSOj solution . . . 72-0 I 3 volumes 95 per cent, alcohol"! „ . , 



(2) „ , 74-9 1 1 volume water J ' * 



The figures for water and alcohol, though rather low, agree approximately 

 with those found by previous investigators ; and the experiments with zinc sulphate 

 solutions confirm Cohu's result,' that an addition of salt which largely affects the 

 conductivity of a solution only slightly increases its dielectric constant. 



An attempt was finally made to determine the constants of common salt and 

 of soda crystals, but owing to the disturbance of the multiple reflections noted 

 above scarcely any interference effects were noticeable. This shows, however, that 

 the constants for these materials approximate more to normal values than the very 

 high figures found for water and alcohol. 



10. On a Familiar Type of Caustic Curves. By J. Laemoe, F.B.S. 



The illustration of the formation of caustics by the reflection of the light of the 

 sun or other point-source from a band of polished metal is in everyday use. But 

 it seems worth while to call attention to the fact that obliquity of the incidence on 

 a cylindrical reflector does not vitiate the experiment as an exact representation of 

 the geometrical caustic. The bright caustic surface formed by reflection from a 

 cylinder is, in fact, always itself cylindrical ; and the caustic curve depicted on any 

 screen placed across it is merely the section of this cylinder formed by the screen. 

 Although, once this proposition is propounded, its reason is plain, yet it does not 

 seem to have occurred to any of the writers on optics. 



It may be shown, also, that when the reflector is a piece of a conical surface 

 the caustic surface is always a conical surface with the same vertex — thus suggest- 

 ing an extension of the theory of actual caustics into spherical geometry, or rather 

 realising in actuality the analogous theory in spherical geometry. More generally, 

 when tlie reflector is such as may be bent flat without stretching, i.e., when it is a 

 piece of a developable surface, the caustic surface is one of the same kind, and a 

 geometrical corjelation may be established between them. 



med. Ann., xlv. 1892, p. 370. 



