Searching for the marriage of Griffith Roberts and Jane Pritchard sometime between 1855 (when the groom would have been about 18 and the bride 25) and 1866 in Caernarvonshire, Wales (where they were both born and resident at the birth of their two known children in 1859 and 1866 ), I haven't found a single marriage in the GRO indices or the North Wales BMD indices (cross-checked against the parish registers that are available on-line) that exactly fits the bill.
However, I have found the marriage of Griffith Roberts to Jane Parry daughter of Henry Prichard in 1858 . The ages difference between the participants matches the age difference in later censuses (Jane 6-7 years older than Griffith). The marriage took place in in Llanllyfni, which is the parish where both children were born, and the bride's residence is the same as the family's residence in the census 3 years later.
If patronymics were still in use of this area of North Wales in 1858,and Jane might be referred to as Jane ap Harry (Jane Parry), then this is a very good match.
So: were patronymics still in use in 1859 in this part of North Wales? And would Jane have been recorded as Jane Parry, even though properly she would have been Jane verch Harry?
Update: @Fortiter very helpfully pointed me to Rowlands, John, and Rowlands, Sheila, eds. The Surnames of Wales for Family Historians and Others, 2nd ed. Baltimore, MD: Genealogical Publishing Company, 1999. ISBN: 0-80631-516-4, where I found:
As late as 1862, the marriage took place in Llangi'an, of Ellen Roberts, daughter of Robert Jones, and such an event, though by then unusual, could be replicated in other parts of rural north and west Wales.
which suggests that the timeframe is possible and also that, by then, verch wasn't always used but that a suffix 's' or prefix 'p' were used to form the second name, even for daughters.
Second update: There's a very good post on the timelines for patronymics in Wales at this Rootsweb thread, confirming that the timeframe is possible.