But when, in spite of all this, we find Mr. J. K. Mackenzie[5] maintaining that the Fuller transformer was non-polar, and further, that the following improvements must be ascribed to Messrs. Gaulard and Gibbs, viz.:— 1. The reduction of the primary and secondary wire-resistance to a minimum. 2. The attainment of the greatest possible coefficient of induction with the lightest apparatus. 3. The symmetrical arrangement of both coils. 4. The proportioning of the coils, so that the weight of metal in each is the same. Seeing this, it must be thought that this gentleman either does or will not, understand the subject. Then if Gaulard has succeeded with his apparatus in obtaining some advantages as proposed in the above-mentioned clauses, Nos. 1 and 2, these advantages can be obtained to a much higher degree with non-polar transformers. This has been proven by Prof. Ferraris.[6] The improvements mentioned under Nos. 3 and 4 are only to be attained with bi-polar transformers after difficult and otherwise disadvantageous arrangements; for instance, the combination of the primary and secondary wires in a common cable, or, when the coils consist of ribbon wire, by the winding of the one inside the other. With non-polar transformers these improvements are already inherent. The Fuller transformer was just as much without poles as two horseshoe magnets are, with their like poles laid together. In all these systems with series connection of the transformers, the intensity of the current in the primary circuit must be held constant in order that it may be possible for the induction apparatus to maintain the secondary electromotive force constant. Notwithstanding this, constancy was not attained, but only one cause of the variations annulled. Another cause of the variations of the difference of potential at the secondary terminals of the coil still remained; this was the loss of potential due to resistance and self-induction, which increased with the load. The electromotive force of the secondary, and therefore of the primary coils, accordingly increases as the current in the secondary decreases. When no secondary current is flowing, the electromotive force in the primary and secondary coils is a maximum. We have consequently this disproportion that the smaller the output of the apparatus the greater the energy consumed. With the secondary circuit open and a constant exciting current, the energy used could be as much as ten times as great as under full load. The disadvantages of this system are apparent; for, putting aside the loss of energy arising from the disproportion between produced and consumed energy, each change of load on the secondary circuit exerted a great influence on the primary circuit, and again on the secondary circuits of the other coils in the main circuit.