One popular, and possible, explanation of these inconsistencies is that Watson was shot while he was bending over at some point during the battle. The bullet could have conceivably passed through his upper leg, then lodged in his shoulder. At various times afterwards, one wound or the other would bother him enough for him to complain about it.
It is generally agreed that Watson met Mary Morstan in September of 1888 (The Sign of the Four) and married her several months later, presumably in the Spring of 1889. Watson, in "The Boscombe Valley Mystery", and Holmes in "The Stockbroker's Clerk", made comments definitely linking this "wife" with The Sign of the Four. In "The Empty House", Watson mentioned that Holmes had learned of his "own sad bereavement". Since Holmes had ostensibly gone over the Falls in April of 1891 and returned in April of 1894, Mary Morstan must have passed away sometime between those dates. In "The Blanched Soldier", which occurred in 1903, Holmes wrote that "Watson had at that time deserted me for a wife". This "wife" was clearly someone other than Mary Morstan, and so it would appear that Watson was married twice. This seems simple enough.
The real difficulty, however, starts in "The Five Orange Pips" when Watson wrote that his "wife was on a visit to her mother's". Since Mary Morstan stated that her mother was dead in The Sign of the Four, and since Watson recorded that "The Five Orange Pips" occurred in September of 1887 (a full year before he met Mary),then this "wife" could not have been her. To make matters even more complex, Watson also mentioned "The Sign of Four", as part of a dialogue with Holmes, in this same story.
And the crux continues in "A Scandal in Bohemia", where Watson clearly recorded that the case started on March 20, 1888 and again referred to being married. Based upon these two stories, one might argue that Watson had a wife before Mary Morstan, but if this is true, then what became of her?
In an attempt to explain these inconsistencies, some Sherlockian scholars re-assigned dates to these stories. But those actions introduced new contradictions, didn't they? As for myself, I'll go with the three-marriage theory.
One interesting theory to resolve this was published by Brad Keefauver in the June 1983 issue of the Baker Street Journal. Mr. Keefauver's theory was that Jabez Wilson was lying to Holmes about the true length of his employment (really 24 weeks rather than 8 weeks). As he also later wrote, "the digging of a tunnel and the copying of all that encyclopaedia material would both fit more comfortably into a twenty-four week span." In other words, Watson correctly recorded the dates of the newspaper and the cardboard sign, but Wilson kept emphasizing an eight-week period. Was this to keep Holmes's fee down?
And speaking of this same case, something that has always bothered me was that 14-year-old girl who did "a bit of simple cooking". How could John Clay have been tunneling under the pavement all that time without her noticing anything amiss? Could she have been an accomplice? What ever happened to her?
In A Study in Scarlet, who was "Mrs. Sawyer"? Like Watson, Jefferson Hope was without kith or kin in England. If he was busily pursuing Drebber and Stangerson while struggling to stay alive as a cab driver, he hardly had the time or opport unity to make any friends. On the other hand, his meager earnings would not allow him to pay someone to help him pull off his rather elaborate ruse. Who did he get to masquerade as an old woman looking for a lost wedding ring? (This crux is a rather minor one, and could be simply dismissed as a "weak spot" in the story. After all, it was ACD's first.)
Part of the fun and enjoyment of Sherlockiana lies in finding, examining, and attempting to explain such inconsistencies in the Canon! This listing of the more obvious cruxes is by no means exhaustive, and certainly can be added to as new theories are developed.
http://www.sherlockian.net/world/cruxes.html
Copyright © Chris Redmond 2005