I was told by a liguistics professor that de-capitalisation happened when the word was sufficiently common to become 'everyday', so to speak. So, perhaps we have 'abelian' rather than 'Abelian' because we use the word a lot. Certainly, we use 'abelian' and 'noetherian' all the time, whereas say, 'Cohen-Macaulay' doesn't appear quite so often.

It seems like this happens more often to 'algebra words' (abelian, noetherian, artinian) than, say 'geometry words' (Euclidian, Riemannian). Could it be that algebraists use the name-words more in 'everyday' speech than geometers? Maybe they even have sloppier attitudes to grammar-pedantry, or like to save their shift keys, who knows?

(Of course, this is a very tentative hypothesis and I expect to see it demolished by counterexamples.)