The following communication has been received from Prof. R. A. Fessenden, from Bermuda, where he is spending the summer. The opening paragraph is an explanation of the unavoidable delay in the transmission of the letter, and is therefore omitted:

SIR--I disagree entirely with the London "Electrician," and consider the sine form of curve to be the best one. It is true that, on account of the presence of iron in the circuits, a rectangular curve will give a slightly greater amount of power in a circuit for a given amount of hysteresis, but this is more than offset by the greater losses from eddy currents and by the greater cost of line and generators and motors for the rectangular curve, if the same amount of loss is to take place in both circuits. Under ordinary circumstances an irregular or rectangular shaped curve would not get very far before it would be modified so as to more closely resemble a sine curve, and one might just as well make the dynamo give the sine curve at once, and so avoid the eddy current and line losses due to the components of higher periodicity in the rectangular curve.

I do not, however, believe that it is necessary, as has been stated by some electricians, to use a surface wound armature to get a sine curve, as a sufficient approximation to that form can be obtained with a properly designed toothed armature.

The experiments of Mr. Scott, of the Westinghouse company, show that in practice, as in theory, the sine curve is the best.

I may say, in this connection, that it does not seem to have been generally noted that the sine curve is a necessity for efficient telegraphy. In January, 1891, I designed and experimented upon the system of multiplex telegraphy which Dr. Pupin has recently rediscovered, and noticed this fact. As a result, a method was devised by which the operator did not make or break the line circuit with his key, but put in circuit a device which automatically sent out sine waves into the line. REGINALD A. FESSENDEN.