Rookie question - who gave the information for the engraving of the stones? I've seen many applications online for WW1 and WW2 and they often have corrections where the information filled out by the family was fixed. Likewise, I've seen a lot of pension applications where the family had facts wrong, especially those done by widows. Could a similar issue be at work here?
I've seen many of the applications and they don't have any info about who submitted them; just the name and unit. In the cemetery where I volunteer, many of the CW stones were obtained by the GAR as a number of the veterans didn't have any local living family. In one case the unit is incorrect and several have the names misspelled. In one case where there was family they added extra engraving to the issued stone; don't see that a lot. Bottom line is it's up for grabs who applied for the stone and how much they actually knew.
Very interesting story and investigation; another one where the group has worked together to solve a mystery. I do love a graveyard mystery.
Edit: I've seen a number of more modern (e.g. WWII) marker applications that had incorrect information. In one case we have a vet who got two markers: one ordered by his mother and another by his brother. Both were delivered to the cemetery but the one ordered by the mother got there first and was installed; the other just sat in a storage shed for decades (until I found it). Turns out both markers are incorrect regarding dates of birth and death. Now, one would think a mother would know at least the birth date but I've got good sources that say otherwise (e.g. the veteran's draft registration, written by him) and his death certificate (confirms the death date for certain). It's one of those things one has to consider when doing genealogical research: many "official" records have errors. It's one reason there's a rule of thumb in genealogy that something shouldn't be considered a fact unless you've got at least three original sources that all agree (more often than not something that never happens). So, in the end you often just have to make the best guess based on what you have and what seems most likely.