1. Since 1955 Harker's work on the control of cockroach activity rhythms by a hormonal clock in the sub-oesophageal ganglion has stood largely unchallenged, but recently Roberts (1966) has questioned several of her claims; in particular he finds it impossible to transfer rhythms by implanting this ganglion.
2. Sub-oesophageal ganglia from rhythmic donors were implanted into 29 headless Periplaneta americana in a variety of ways and three different actographs used. In 2 cases the hosts showed bursts of activity for 2 or 3 days after implantation, roughly coincident with the donor's previous rhythm; a further 8 implanted animals showed rather uncertain signs of an induced rhythm, but the remaimng 19 were all apparently arrhythmic.
3. Cauterization of the neurosecretory cells of the sub-oesophageal ganglion in situ showed that cockroaches can remain rhythmic in the absence of the cells described by Harker (1960c), and probably in the absence of all cells in this ganglion which are stained by paraldehyde-fuchsin.
4. Implantation of abdominal ganglia plus their respective neurohaemal organs (Brady & Maddrell, 1967) did not elicit rhythms in eight headless hosts.
5. Cutting the circum-oesophageal commissures, or post-sub-oesophageal ganglion connectives, like beheading, appears to interfere seriously with the rhythmic expression of activity; one animal remained apparently rhythmic after this operation, however.
6. Harker's proposal for a second clock (1960b requires that phase-shifts of less than 5 hr. shall be completed within a single activity cycle; this did not occur in the twelve cases observed, phase-shifting being gradual and taking several days before reaching the entrained steady-state.
7. It is suggested that most of the divergent results of Harker, Roberts and Brady could be plausibly correlated if cockroaches have an electrical pace-maker in the brain co-ordinating rather ephemeral neuro-endocrine rhythms in the nerve-cord ganglia.