On December 10, 1946, Martin Gardner in Chicago wrote letters to Marshall D. Smith, Richard W. Hood (son of and successor to Edwin C. Hood, founder of H. C. Evans & Company, the Chicago based gambling supplier since 1892) and the Canadian Copyright office, asking all of them specific questions about S. W. Erdnase and his book. All responded promptly and only the Canadian copyright office yielded no information, other than the fact that they could find no record of copyright there. Smith responded just two days later and in his reply letter he wrote: I did the drawings for Mr. Erdnase whose name I had forgotten. When Gardner met Smith the very next day, Gardners notes tell us: Before I [Gardner] mentioned Andrews as the name, he said that Erdnase didnt sound right, and he recalled it as a name with a W. When I said Andrews his face lighted up and he was sure that was it. Does not recall first name or initials. I think it worth noting that Smith did not independently recall the name as Andrews, though he strongly supported Gardners suggestion. Gardners interview with Smith and his subsequent correspondence yielded quite a bit of specific information regarding the books author, including a detailed physical description and the fact that he was somehow related to Louis Dalrymple, the famous political cartoonist of the period. He also recalled that he made pencil sketches of the authors hands, then took them home to ink them in after the author had OKd each sketch. He thought the job took him about two weeks, though he had specific memories of only their initial meeting
Just a month later, on July 16th, 1947, Gardner wrote the Librarian of Congress for the first time about the book. In that letter he says: The authors real name was James Andrews. He obtained the pseudonym of S. W. Erdnase by spelling his real name backwards, including the last two letters of James.
In his reply some two months later (March 17th, 1947), Robert C. Gooch, Chief of the General Reference and Bibliography Division, after supplying the bibliographic information Gardner requested, writes: We are very interested to note that you have discovered evidence that this authors real name is James Andrews. Our Processing Department would be pleased to learn in what source this information may be found, in order to complete its records. In his detailed response of March 20, 1947, Gardner writes: Regarding the authors real name: In my research on Erdnase I located M. D. Smith, the artist who did the illustrations. He lives in Chicago, a hale and hearty man of about 80 [in fact, he 74 at the time -rh]. He remembered Erdnases real name (I.e. James Andrews). With this as a lead, I found a magazine article by James Andrews in Harpers Weekly, June 26, 1909, titled Confessions of a Fakir, which contains intrinsic material that establishes it beyond doubt as by the same author as the book on gambling methods. This article was reprinted in Conjurers Magazine in August 1949. Just two months later, in October 1949, Gardner found articles from 1905 detailing the lurid life and death of card cheat Milton Franklin Andrews, who had been described to him as Erdnase (without revealing his name) by Philadelphia magician, E. L. Pratt. Within a short period of time, Gardner abandoned the James Andrews theory in favor of Milton Franklin Andrews.
What surprised me in Gardners correspondence was the claim that he was led to the James Andrews theory by Marshall Smiths recollection. He met Smith in December 1946 and makes this claim in March 1947, though mentions the James Andrews name just one month after meeting Smith. He made no mention of James Andrews in his article THE MYSTERY OF ERDNASE published in the SAM Convention program in May 1947. James Andrews is mentioned in Vincent Starretts weekly Books Alive column in the Chicago Sunday Tribune of June 15th, 1947: For nearly half a century the identity of Erdnase remained a mystery; then the ingenious Mr. Gardner read the name backwards and produced E. S. Andrews. But who was E. S. Andrews? A later discovery by Mr. Gardner revealed him as James Andrews; the initials obtained by spelling the name in reverse were the last two letters of James. This final revelation came too late for inclusion in Mr. Gardners article, The Mystery of Erdnase, and were revealed to me in a letter supplementing the printed revelation The same article mentions Smith, but without crediting him with this revelation. It does credit Smith with the Louis Dalrymple clue, noting that Dalrymple was then  a cartoonist and comic artist for the Chicago Tribune. (Incidentally, Smith acknowledged receiving a copy of the Tribune article from Gardner in his letter of June 24, 1947).
Alas, Gardners own recollection of this episode is now pretty dim (he is more than a decade older than Smith was back then and it was 55 years ago!). He now thinks it likely that he first found the article in Harpers Weekly, then asked Smith about the name James Andrews and got some kind of encouragement, though this is, of course, not what Gardner wrote to the Library of Congress at the time. And why did he omit the reference to James in the SAM Program? Surely not, as the Tribune article states, because he obtained it too late for inclusion. He had the information in January, the convention wasnt till May
Some of you may recall that I was once enthusiastic about a James Andrews candidate myself, specifically, James DeWitt Andrews, a Chicago attorney and writer of legal treatises. I remain interested in James DeWitt Andrews, but in trying to link him to Dalrymple, I stumbled across Edwin Sumner Andrews, whom I consider a more likely fit on circumstantial grounds. The most intriguing response to the MAGIC article (December 1999) I wrote on this topic (which included considerable information on James DeWitt Andrews) came from reader Michael DeMarco. He found the circumstantial case I made for JDA sufficiently compelling to search the first edition title page (which seems to be the Rosetta stone of this mystery) for the other missing letters of his name. Sure enough, there they are: the first letters of each line of the inverted pyramid subtitle are JAM DEWTT, missing only the letter I (no, they are not in that order!).