This is a minor note about a grammatical anomaly.
Reading Leonard Palmer’s classic book The Latin Language (1954, reprinted in 1988), I came across a point about a fossilised form of Early Latin. It turns out that the -as ending of Pater familias (“Father of the family/household”) is actually an archaic genitive (p.97).
Probably this explains why Familia either does not decline in this set phrase (for instance, it has an attested genitive form of Patris familias, and of Patrem familias as accusative), or, when it does, it declines as any other first declension feminine noun (i.e. as -ae, etc.); and why it does not decline the way Latin nouns ending in -as in the nominative singular tend to do (that is, as third declension nouns; in this case, as *Patris familiaris in the genitive).
Pater familiae is also attested. It appears for example in Caesar’s De Bello Civili:
Quibus rebus accidit, ut pauci milites patresque familiae, qui aut gratia aut misericordia valerent aut naves adnare possent, recepti in Siciliam incolumes pervenirent.
(From these things, it came about that a few soldiers and patres familiae, who survived thanks to their influence or mercy granted to them, or who were able to swim out to the ships, were received unharmed and came into Sicily; DBC II.44)
There are a few other places where this archaic -as genitive survives, though in much older fragments. For example, in Livius Andronicus’ (ca.284-205 B.C.) Latin translation of the Odyssey, we find this description of the muses:
Nam diva Monetas filia docuit…
(For the divine daughter of Moneta has taught…)
It would be interesting to see whether this anomalous form results in confusion in later Latin. I wouldn’t be so surprised to find a medieval author declining it as, e.g., *Patris familiaris in the genitive (*Patrem familiarem in the accusative, etc.). I will keep my eye out.
Palmer, Leonard. The Latin Language. Oklahoma: University of Oklahoma Press, 1988.