On pg. 242 I give a complete paradigm of mare as an exemplum for the neuter i-stems, but Erica Bexley of Cornell has called my attention to the fact that the expected genitive plural marium is never attested in all of Classical Latinity. I couldn't believe it at first, but it is true. The absence of the the oblique plural forms of mare was noted by the Grammarians. The 4th-century CE Charisius (p. 38.6 Barwick) wrote:
maria tamen quamvis dicantur pluraliter, attamen ne marium nec maribus dicemus.
(Followed by Ars Bobiensis (M. De Nonno 1982: 29, 22): quamuis maria dicitur nec marium nec maribus... accipimus and [Augustinus] Regulae: mare maria: sed pluraliter tres casus habet tantum, nominatiuum accusatiuum et uocatiuum, genitiuum datiuum et ablatiuum non habet: non enim dicimus in genitiuo horum marium aut in datiuo his maribus aut in ablatiuo ab his maribus.)
Priscian (2.351 Keil) gave an actual example of the supposedly nonexistent ablative plural maribus from Caesar and quote a line of Naevius with the genitive plural marum, not marium:
et sciendum tamen, quod rarissime haec, quae in solam i finiunt ablatiuum, syncopam patiuntur i per genetiuum pluralem. inueni marum pro marium, qui tamen in rarost usu genetiuus, apud Naeuium in carmine belli Punici (Blänsdorf 9):
senex fretus pietatei deum adlocutus
summi deum regis fratrem Neptunum
pro marium. eius ablatiuum Caesar in V belli Gallici (5.1) ponit: paulo latiores quam quibus in reliquis utimur maribus.
Basically the word seems to have been defective in the plural. The form marum is surprising since we are confident that this word has been an i-stem for a very long time having an exact morphological match in Old Irish muir 'sea' < *mori. It must be analogical.
It's unbelievable that I could have lived 45 years and written a book on Latin historical grammar without knowing this!