Manchu words like giranggi ‘bone’ or senggi ‘blood’ are traditionally classified as part of a nominal class system inherited from the Proto-Tungusic parent language. This nominal class system has obscure origins and it is unproductive in the historical languages (Manchu included). Manchu giranggi and senggi contain the class suffix +nggi which in historical terms cannot be easily reconciled with the so-called “collective” suffixes +ksa and +kta in Core Tungusic languages. In this contribution we argue that the class suffix +nggi is the result of the reanalysis *…V/n+g.i ⇒ …V+nggi whereby nasal nouns are reinterpreted as vowel nouns. Common Tungusic “collectives” *+kta and *+ksa are secondary formations that were created after Manchu had branched off. The general assumption is made that Manchuric (a.k.a. Jurchenic) serves best to improve our understanding of the prehistory of the Tungusic languages when it is seen as the conservative member of the family instead of the innovative one as usual.