A longer version of this comparison was noted by Clark (2011:318-319), who proposed an ambiguous reconstruction *(ku(r,l)u)ku(r,l)u ‘pigeon sp.’. However, the forms cited here support an unambiguous reduplication (Tongan*r > l rather than the more usual zero is rare, but recurrent, and some doublets are distinguished by these reflexes, as with POc *rua > Tonganua ‘two’, but lua ‘two (in some combination forms)’.