Name of amateur who gave a new proof of the Ramanujan-Nagell theorem? In an article by George Johnson in the New York Times back in 1999, it says that an amateur mathematician from India once sent Ian Stewart a proof of the Ramanujan-Nagell theorem that the Diophantine equation $x^2 + 7 = 2^n$ is solvable if and only if $n = 3, 4, 5, 7, 15$.  The proof "was badly typed on strange paper and cast in an idiosyncratic style that would have given any journal editor the impression that the writer was a crank."  However, it was correct, and after getting some help cleaning it up, the man published the proof.
To me, this is an inspiring story, and I would like to know the name of this man and to see the paper.  I asked Ian Stewart but he said that he remembers the incident but not the identity of the man in question.  I would try asking George Johnson but I am not sure how to contact him.  I searched MathSciNet but was not able to guess which paper it was.
Does anyone know more details?
 A: I'm going to argue that the story is apocryphal; there was no such proof by an amateur.  Here's why
The linked story is not about the amateur and the proof; it's about mathematical cranks.  The amateur is not named; all the article says is

Several years ago Dr. Stewart heard from a man -- in India again --
  who had found a new, simpler proof for an obscure, pointless theorem
  in number theory written by Ramanujan and a collaborator. According to
  the Ramanujan-Nagell theorem, the only numbers one can square and add
  7 to, yielding an answer that is a power of 2, are 1, 3, 5, 11 and
  181. For example, squaring 3 and adding 7 gives 16, which is the fourth power (the square of the square) of 2.
Dr. Stewart was surprised to realize that the proof was correct, but
  it was badly typed on strange paper and cast in an idiosyncratic style
  that would have given any journal editor the impression that the
  writer was a crank. Dr. Stewart advised the writer to find an Indian
  number theorist who could teach him how to present a proper paper.
  Several years later the result was published, and soon after came
  another publication from the same man. ''It is worth reading these
  things occasionally,'' Dr. Stewart said.

A search on MathSciNet for any article between 1998 and 2003 containing both the words Ramanujan and Nagell turns up only 19 hits.  None are the article in question.
A Google Scholar search with the same parameters turns up only 4 pages of hits.  None are plausibly an article of a proof by one or two authors, (at least one from India)
The 2002 survey article Relevance of Srinivasa Ramanujan at the dawn of the new millennium by KS Rao talks about the Ramanujan-Nagell equation does not mention a new proof - one might expect that Rao would have known about this.

So what did happen?  Where did the story come from?  One could imagine that Stewart had some conversation with Johnson about mathematical cranks, and misremembered  a story, which Johnson did not do enough checking to correct.  (side note: Johnson writes regularly about science but has only written a couple of math articles)
