Experimental grafting between cloves is very difficult with shoots, and so far has proved impossible with roots. Use has therefore been made of naturally occurring grafts in the study of sudden‐death disease. Volunteer seedlings often grow up closely adpressed to old trees. If the old tree dies from sudden‐death disease, the pole† usually survives, but occasionally it dies almost simultaneously with the old tree. The poles discussed invariably had an independent root system, but those which died were found to have their roots grafted to those of the old tree whilst poles which survived, although closely adpressed to the old tree, had no organic connexion with it. These observations cannot be reconciled with any but a pathogenic hypothesis as to the nature of the sudden‐death