Early experiments
The early attempts to signal without wires fall into three categories. The methods employed successively were signaling by conduction, by induction, and by radiation, the latter being the successful method in use today. Interesting though they are, the first two methods have but little practical bearing on the later method, although they did play an important part in its evolution. For this reason we shall refer to them as briefly as possible.
All three methods use the ground as part of the circuit. That it was possible to complete an electrical circuit through the ground appears to have first been discovered by Winkler, of Leipzig. In 1746 he discharged Leyden jars through an insulated wire laid along the bank of the River Pleiss, the waters of which formed the return half of the circuit. Later in the same year he successfully transmitted over a distance of two miles, using the ground as a return circuit. Subsequently, there were numerous other instances in which the ground return was employed, as in I747 when an Englishman, Dr. (afterwards Sir William) Watson transmitted an electric current over a single wire 2,800 ft. in length and followed this by transmitting over a distance of two miles, in each case using the ground as a return.
(page 423)
[Editor's note: I did the illustration below; it is not part of the original article but it shows its key points]
Three methods exist to communicate wirelessly: conduction, induction, and radiation. (Transmitting by optical means, be it the infrared of a television remote or the visible light of a laser (internal link), falls under radiation, since it also employs radiant energy.) Radiation is how nearly all wireless has been conducted since Marconi.
Click here for a selection from Weisman's RF & Wireless. Easy to read, affordable book on wireless basics. (12 pages, 72K in .pdf)