The Rest of the Story
The legend of Nathan B. Stubblefield, as presented above, is essentially correct; Stubblefield did indeed exist, he did demonstrate a wireless telephone that transmitted voices as early as 1902 (if not ten years earlier), he did receive Canadian and U.S. patents for a wireless telephone, and he did indeed go into seclusion only to eventually die of starvation in poverty.
What is not told of by the legend is that Stubblefield, in mid 1902, had agreed to a commercial use of his invention and was named the 'director' of the "Wireless Telephone Company of America," but held no office; he received stock in the new company in exchange for the patent rights to his device. His demonstration in Philadelphia was part of the promotion of the device, to help sell stock in the new company. Shortly after the Philadelphia demonstrations, however, Stubblefield withdrew from the company; while the exact reasons for this decision are unclear, there were several events that may have prompted the choice.
After the Philadelphia demonstrations, a fellow inventor and experimenter from that city joined the new company. His name was A. Frederick Collins, and he had separately developed a wireless telephone system that only had superficial differences from Stubblefield's system. Unlike Stubblefield, however, Collins was well-published in the scientific journals of the time, so his name carried more weight, promotion-wise; after Collins joined the company, Stubblefield and his device lost a great deal of his prominence in the company’s advertisements. It has also been alleged by some that Collins was part of a group that conspired to steal Stubblefield's designs, but this cannot be proven.
Another strain between Stubblefield and the company was revealed by a letter that Stubblefield wrote to the secretary which clearly indicated that he felt he was somehow being swindled by them, and that all the company was actually interested in was selling stock. This seems likely, as the company never applied for the patents for either Stubblefield’s or Collins’ telephones, nor did the company make any actual improvements on the telephone systems; and in 1913 some of its officials, including Collins, were convicted of mail fraud.
It was after his break with the company that Stubblefield developed a second version of his system, different enough from the first that he could legally patent it in his own name, which he did in 1908. Neither the design promoted by the Wireless Telephone Company of America nor the design patented by Stubblefield himself were ever commercially successful, which is likely what finally drove the inventor into seclusion and poverty.
But why was it not commercially successful? The answer depends on one important question...

Exactly What Did Stubblefield Invent?
The claims for Stubblefield's accomplishments were summed up best by L.J. Horten in 1937 when he said: "...Nathan B. Stubblefield was the first to discover, invent, manufacture, and demonstrate equipment for broadcasting and receiving the human voice and music by wireless. He invented the radio". This represents the first error generally made in regards to Stubblefield and his work: there are many ways to transmit information, of which radio is just one.
The principles behind Stubblefield's devices were worked out in 1970 by Elliot Sivowitch based on an examination of the 1908 patents (U.S. Patent No. 887,357, dated May 12, 1908, Serial No. 366,544, dated April 5, 1907. Canadian Patent No. 114,737, dated October 20, 1908), and verified in 1971 by Thomas Hoffer based on a second examination of the patents, as well as an examination of the collections of papers and photos related to Stubblefield held at the University of Kentucky in Lexington and at the Chamber of Commerce in Murray, and a personal interview with Stubblefield's son Bernard, the only one of Stubblefield's children to have been allowed to work with and on his father's actual wireless telephones.
Both of Stubblefield's designs -- the earlier one owned and promoted by the Wireless Telephone company, and the later one patented in 1908 by the inventor himself -- relied on the use of the principle of induction to transmit sound, the first through the ground from metal rod to metal rod, and the second through the air from antenna to antenna. Induction refers to the fact that a change in a magnetic field can create a current of electricity, and a current of electricity can create a change in a magnetic field; thus a current in one wire can produce a current in another wire, even at a distance, and this is indeed the same principle by which radio works.
Previous to Stubblefield’s experiments, other experimenters had theorized about the possibility of creating a wireless communication using this principle. Samuel F.B. Morse, inventor of the telegraph and the "Morse Code", had conducted experiments along these lines as early as 1842; and after the introduction of the telephone and the mass laying out of underground telephone cables, it had been noticed that voices would sometimes cross over from one wire to another using the ground in between as the conductor. The earliest documented occurrence of this I have is in 1877, when a concert being "broadcast" to telephones in Saratoga Springs, New York, from New York City was also heard accidentally in both Providence and Boston because of electrical leakages between adjacent underground telephone wires.
But induction transmission does differ from radio transmission. With induction transmissions, most of the energy of the transmission is confined to the vicinity of the transmitter wire; with radio transmissions, almost all of the energy of the transmission is actually transmitted. What this means in simple terms is that radio transmits much farther than induction... in Stubblefield’s telephones, the design allowed for a maximum transmission range of only a little less than three miles, whereas even the earliest radio transmitters had a range of thousands of miles.

Other Transmissions
Previous to the creation of Stubblefield’s wireless telephones, induction transmission was already being demonstrated as a form of wireless communication by Professor Amos Dolbear of Tufts College, Massachusetts. As early as March 1882, Dolbear had a telephone set-up that used phones grounded by metal rods poked into the earth; his transmission range was a little less than a mile, and he received a patent for it (U.S. Patent No. 350,299). Ten years later a man by the name of John Stone, funded by AT&T, made successful voice transmissions over telephones to ships at sea in 1892; he, too, was using a form of induction rather than radio broadcast to accomplish the feat.
In 1895, Guglielmo Marconi had developed the first wireless system using radio as the transmission; this first system was just a Morse code telegraph, but by 1900 Reginald A. Fessenden, an American physicist, made the first recorded transmission of voices by radio. This was two years before Stubblefield’s first public demonstration, which is the reason his supporters stress the fact he had shown his wireless telephone to a man named Wells in 1892. By the time Stubblefield patented the second design of his wireless system in 1908, it was two years after Fessenden had first broadcast phonograph music by radio in 1906.
So, unfortunately for the legend of Nathan B. Stubblefield’s wireless, he neither invented radio, nor made the first voice transmissions by wireless. However, it has been pointed out that Stubblefield’s son, Bernard, may make claim to something almost as important in the history of broadcasting: all evidence supports the fact that Bernard was the first entertainer to play live music to a broadcast audience.