Many thanks, folks!
Especially Jenn for the actual death records of James and William. It seems clear that there is no positive evidence in Tasmania that William who died there in 1895 came from Somerset.
Christine's find of James producing children in South Australia and a marriage to Mary Symes in Chard in 1858 is interesting.
IF the James who was transported in 1843 is the James who was married to Elizabeth and lived in Winsham in 1841, then IF he returned at the end of his sentence, he would have found his wife cohabiting with and having children by another man, a Henry Hillard. In his shoes I would have found another "wife" and cleared off back to Australia!
Mebe - I think the link between a William Grimstead born 1831 in England who died in Launceston in 1895 and a William born in Winsham, Somerset in 1831, is because he is simply the only one that fits!
In 1841 there were only two William Grimsteads in England who were born in 1831, and both of them were born in Winsham in Somerset. (There may of course have been others who were mistranscribed, but I haven't found any!)
These two Williams have been tracked to registered deaths in England, so William of Launceston is unlikely to be either of them.
But of course, once a family link appears on the IGI, especially in an Ancestral File or Pedigree Resource file, it is taken as fact, and self-propagates.....
I have a similar case in England in Bedfordshire - which is almost totally covered by extractions on the IGI up to 1812. There is only one baptism of a William Armstrong in 1792 in Bedfordshire that "sort-of" fits a census trail elsewhere in the same county, ergo this must be the man! But no-one looked at burials, so they didn't discover that the baptised William died at the age of 15....so simply could not have married twice and produced two families.
This error is in the tree of almost every descendant of the real William, who was actually never baptised! Or if he was, it was as an adult after 1812 in a parish the IGI didn't cover after that year....