When is a son not a son?

No, not an existential question, just a comment on the sometimes idiosyncratic indexing of census "families" on ancestry.

The 1881 census for Maryhill, Lanarkshire includes the military Barracks there.
In residence were William PAXTON, wife Mary PAXTON (formerly RUNCIMAN, nee DOUGAN/DUGGAN), John, Frances and Norman, her children from her previous marriage to George RUNCIMAN, and Mary, her daughter with William.

How do (John,) Frances and Norman appear on ancestry?

As children of William and  Bridget REYNOLDS:
William Reynolds 36 Head Soldier b(orn) East Indies
Bridget Reynolds 33 Wife b. Ireland
John Runciman 15 Soldier b. Ireland
Francis Runciman 8 Daughter Scholar b. Gibraltar
Norman Runciman 4 Son b. Malta

FindMyPast shows each of the family members individually (ie not listed as a family, but reported as head, wife, son..., my regrouping):
PAXTON, William Head Married M(ale) 34 1847 Soldier (born) England
PAXTON, Mary Wife Married F 31 1850 Lemerick
RUNCIMAN, John M 15 1866 Soldier Ireland
RUNCIMAN, Francis Daughter F 8 1873 Scholar Malta
RUNCIMAN, Norman Son M 4 1877 Malta
PAXTON, Mary Daughter F 2 1879 Hongkong

I confess to not having checked the original record, but do have sufficient evidence from other records for my conclusions.
A number of family trees on ancestry have come to a dead end on their ancestors as a result of not checking further, and assuming William & Bridget REYNOLDS are their ancestors.

For those interested, this family is included in the RUNCIMAN One Name Study.