Sound is a very complicated thing. It can be reflected in much the same way as light, and I suppose most school-boys know that if a concave mirror is at one end of a room and a similar mirror at the other with a watch hanging at its focus, the watch cannot be heard by an observer walking across the room, yet as soon as he places his ear at the focus of the other mirror he will hear the tick clearly, showing that sound is easily reflected. Everybody who has heard an echo should know this. Sound travels also very slowly, and there is plenty of time for wind and different mechanical scraping effects to spoil the purity and partially absorb its delicacies. [Illustration: Fig. 1. Sound can be focussed by a concave reflector and by this means the tick of a watch can be heard at a distance. The watch and the end of the wooden rod are placed at the focus of the reflector.] Remember that if I am addressing a man by wireless who is one hundred miles away, someone who is listening on a telephone will hear my voice before I am heard at the end of the big hall where I am speaking, because the velocity of sound is only 1100 feet per second, and wireless, like light, travels much faster. Sound can also be actually refracted. Just as the old-fashioned jeweller used a globe of water to concentrate the light upon his work, so will a collodion balloon filled with carbon dioxide, the ordinary gas product of average combustion, act as a lens for sound, which can be actually focussed by these means. This exemplifies the complication of our subject, and indicates that the heavy diaphragm and other details of the loud speaker must produce serious distortion. Let us be honest at once. We can only hear such distant places as America by the grace of heaven. Even Sunlight can tune sweet song into vague cracklings. Until true tuning can be obtained we are largely at the mercy of the reproducing instrument, which too often exaggerates every fault and gives the impression that wireless and music are in no way related. No loud speaker of to-day really produces voice and song which sound exactly like voice and song. It all too much resembles a bad gramophone, but without the advantage of the user having the choice of the music.