Is there support for claiming -gn- was pronounced as /ŋ/ in classical Latin?

Question

According to what I have learned, -gn- was commonly pronounced /ŋn/, e.g. [ˈmaŋ.nʊs] (magnus). However, this excerpt from Encyclopædia Britannica had me wondering:

The sound represented by ng (pronounced as in English sing and represented in the IPA by /ŋ/), written ng or gn, may not have had phonemic status (in spite of the pair annus/agnus ‘year’/‘lamb,’ in which /ŋ/ may be regarded as a positional variant of /g/).

Could it be that -gn- was in fact pronounced merely /ŋ/, so that you would get annus–agnus as [ˈan.nʊs]–[ˈaŋ[ː]ʊs]? ([ː] added due to my uncertainty)? Were this the case, then I suppose one would be unable to tell the difference between e.g. *angus and agnus, so I am now left perplexed.

───

P. S.: I am unsure whether or not phonetic specificity should be added to the tags. (If so, velar and nasal should be added.) I thought it wise to inform of this.

Answer 1

I think the wording of the Encyclopædia Britannica article is unclear, leading to misunderstanding. I don't think they mean that Latin "gn" could represent [ŋ(ː)] rather than [ŋn], but I'll admit the wording is confusing.

The point they're trying to make (I think) is that on the face of it, if [ŋ] can appear before a velar consonant and [n], it looks like it should be regarded as a separate phoneme, but maybe in "ng", pronounced [ŋg], it can be analysed phonemically as /ng/, but in "gn", pronounced [ŋn], it can be analysed phonemically as /gn/.