Filip De Decker alerts me to a confusing formulation on pg. 316 where I identify a subclass of -ti-stems with o-grade including frons, frontis; fons, fontis; mons, montis and pons, pontis. Synchronically this is of course true, but in the case of pons the root is *pent- (cf. Gk. πόντ-ος ‘sea’ and PDE find) and hence the morpheme boundary was originally after the t (pont-) and not before it (pon-t-).