Existing information on the reduction of spurious signals from instrument cables is reviewed. A tentative theory is formulated to account for the known data. Subsequent experiments confirm this theory and allow the formulation of a detailed mechanism of noise generation in cables. The noise may be eliminated for practical purposes (reduced by a factor of 500 or more in some cases) by having continuous electrical conduction between both surfaces of the insulating dielectric and the adjacent conductors. A simple procedure is outlined to make short lengths of this cable for laboratory use, and an industrially applicable method is suggested. There appears to be no limit on how small the diameter of such cables could be made.