Even a simplified scheme of the various neuronal pathways potentially relevant in the generation of essential tremor (figure, A) indicates that didactically attempting to attribute its origin to a single lesion in a single nucleus by means of simple recording is likely to be difficult, and that it may be more realistic to appreciate that tremors probably arise through interactions between central and peripheral neuromuscular systems. The finding of a positive correlation (quantified mathematically as coherence) between a 5–10 Hz electromyographic (EMG) modulation and a component of the EEG or MEG simply indicates the presence of two oscillations with similar frequency and a fixed-phase relation. They are thus likely to originate from the same source but, as Hellwig and colleagues point out, the cortical manifestation could simply represent spread of the modulation along neuronal pathways from other unrecorded structures. Even peripheral sources of oscillation may be reflected back to the central nervous system (CNS) in this manner via sensory pathways. Although phase information, an integral part of any coherence study, gives conduction times for the pathways connecting the coherent oscillations, it cannot easily distinguish the direction of spread because the data are circular, and there is great potential for bidirectional spread of oscillations, which will have complex effects on phase.

In fact, the bulk of the evidence from recordings in animal models and from stereotaxic work in patients points not to the cortex but to two other potential sites for the origin of essential tremor. First, an oscillation similar to essential tremor has been shown to arise through olivocerebellar circuits. 3 Oscillatory activity in normal animals may result from synchronisation via cross-linking between olivary neurons mediated by γ-aminobutyric acid. 3 Perhaps in essential tremor, an amplification of this tendency, whether by a channelopathy or a specific structural synaptic defect, might result in an oscillation that becomes distributed more widely through the CNS and ultimately to the periphery as tremor. Second, the thalamus has a tendency to generate a 6 Hz range tremor through the behaviour of low threshold Ca2+ conductances; 4 the loss of normal inputs from the basal ganglia and cerebellum may release this tendency and has been proposed to result, respectively, in Parkinson’s disease and cerebellar intention tremors. 5 Essential tremor might arise through some other loss of input or from an abnormality of the oscillatory tendency itself. Even if tremor does not arise from the thalamus, reciprocal loop connections between thalamic nuclei and the medial thalamus may act to amplify transmitted oscillations. 4 CNS loops may well turn out to be more important than CNS nuclei in generating oscillatory activity. For example, experimental lesions in various sites that might interrupt CNS loops are effective in ameliorating oscillations resembling those of essential tremor. 2 The ventro-