Family Lore: Nice to hear, but mostly disproven but with some credence. The fact his mother's maiden name was "Charles," may have been the cause of the problem.
Other records as identified:
Marriage "Certificate" records seldom show the names of the parents, although some County forms did include that information. What may contain the names of the parents are the marriage "license application" form...IF they were retained by the county.
The US Census forms are FILLED with errors, ages, (month/year of birth in the 1900 census), major spelling errors of last names, etc. etc. You can use the census as a good reference tool, but you have to filter out errors with other known facts. The errors were caused by many reasons: the person providing the information was wrong, did not know, guessed. A lot of data is provided by persons other than the actual person. The enumerator wrote down the wrong information, they later transposed it onto the forms wrong. A lot of errors in spelling were made due to language, accents, etc. Some enumerators appeared to be very good, educated, good penmanship, others... not so.
Death Certificates: They were only as accurate as the knowledge of the informant, the person providing the information. I've seen information very wrong on these as the informant, even the spouse, did not know the "truth."
The World War I Draft Registration Card image database. It has the image of the actual form used to register. In the last registration, June 1918, I've found about half of the year of birth to be wrong by one year.
The Oregon State Archives: http://genealogy.state.or.us/
They do identify one of his marriages. Others with same last name had filed for delayed birth certificate. A person did NOT have to file in the same county where they were born. So, one of these "might" be related.
The California Death Index:
Thomas Franklin Mercier, born 19 Feb 1881 in Oregon, died 21 Aug 1955 in LA COunty. His Social Security Number was 700-12-1999.
A number beginning with "700-728" was issued by the Railroad Board, which means he had worked for the railroad. This opens up two record sources for you.
1. If he worked for the railroad AFTER 1936, there could be records about him:http://www.rrb.gov/mep/genealogy.asp
2. When he applied for his SSN, he would have completed the Social Security Administration Form SS-5. That form would have included his parents names, among other information. You can order a copy of the actual image of that form and it would identify his parents, at least who "HE," identified himself, as being his parents:http://genealogy.about.com/od/online_records/a/ss5_request.h...