Numbers do matter; they mean that periodicals are full run and
complete, that minor periodicals are included, that successive editions have been duly obtained.
But not numbers alone. I should think that many
numismatics books are not properly distributed, are not announced outside their own region, are in languages (such as Albanian) that many librarians don't read, are published in very small editions, and so on. Even small museums' catalogues. There are, besides,
auction catalogues, especially the 19th-century ones, which are very useful if you need to track the 'life' of a
rare or otherwise very important coin.
I remember once, when a student newspaper published a short article saying that the university
library had reached two million books (that was years and years ago), that someone at a private university retorted that unfortunately that included half a million annuals printed by the State, full of agricultural statistics, and the like, long since digested into permanent tomes (and now, mercifully, digitized) at the state capitol, which also
had the 'originals'.
I should not be surprised, either, if famous coin collectors who could afford to not be dealers or professors were not also bibliophiles, not least in
rare and magnificent numismatic works; such items do not much boost the total number of books accessioned.
Pat L.