I think the 1841 census really says
Wm Jeffery "do" (ditto), i.e. Barrett
and so Mark, also a "do", would also be recorded as a Barrett.
The pattern of a surname for a middle name often indicates that the parents were not married, the child was registered/baptised in the mother's surname, and the father's surname was included as a middle name. It wasn't uncommon for the child to use the father's surname later in life.
Of course, since his name appears the other way around in 1851 -- William B Jeffery -- it could be that the mother was Jeffery and the father Barrett, rather than vice versa.
Just for info too, the 3-yr-old child in 1851 who appears to be recorded as Archelaus Barrett Jane was registered with surname Jeffery -- shown as Archelaus Bennett Jane Jeffery at FreeBMD -- in Sep Q 1847 at Redruth. That's quite the tangle of surnames.
It might be worthwhile to get that birth certificate to see who the mother was and whether a father was recorded.
(Ancestry still shows the mistranscription "Archelsus" but the image there shows Archelaus and also Bennett clearly -- the Bennett is undoubtedly a mistranscription from the original document, but it is what is needed for ordering a certificate.)
Also, Ann Jane Jeffery's birth was registered Mar Q 1842 at Redruth, and Mark's Sep 1840 at Redruth. Since it really does seem they were siblings, I'd get Mark's birth certificate. Mark's age (10 months) was stated very accurately in 1841, so if William's was also, he would have been born between April 1838 and March 1839, and there is only one possible match in Redruth: Jun Q 1839, for a birth late in Mar Q 1839.
I'd certainly start with your William's marriage certificate, as suggested, to see whether a father's name is given!