The power of Parkinson tremor is expressed in terms of possibly changed frequency response functions between relevant variables in the neuromuscular system. The derivation starts out from a linear loopless equivalent model of mechanisms for general tremor generation. Hypothetical changes in this model from the substrate of the disease are indicated, and possible ones are inferred from literature about experiments on patients. The result indicates that in these patients tremor appears to have been generated in loops, which did not include the brain area which in surgery usually is inactivated. For some patients in the literature, these loops could involve muscle length receptors, the static sensitivity of which may have been enlarged by pathological brain activity.