These instruments were, therefore, selected for the further experiments, which consisted of using local extensions in Paris and London.  The wires were in the first instance extended at the Paris end to the Observatory through an exchange at the Avenue des Gobelines.  The length of this local line is 7 kms.  The wires are guttapercha-covered, placed underground, and not suitable for giving the best results.

The results were, however, fairly satisfactory.  The wires were extended to the Treasury in London by means of the ordinary underground system.  The distance is about two miles, and although the volume of sound and clearness of articulation were perceptibly reduced by these additions to the circuit, conversation was quite practicable.

Further trials were also made from the Avenue des Gobelines on underground wires of five kilometers long, and also with some renters in Paris with fairly satisfactory results.  The selected telephones were equally efficient in all cases, which proves that to maintain easy conversation when the trunk wires are extended to local points it is only necessary that the local lines shall be of a standard not lower than that of the trunk line.  The experiments also confirm the conclusion that long-distance speaking is solely a question of the circuit and its environments, and not one of apparatus.  The instruments finally selected for actual work were Gower-Bell for London and Roulez for Paris.

3.  The results are certainly most satisfactory.  There is no circuit in or out of London on which speech is more perfect than it is between London and Paris.  In fact, it is better than I anticipated, and better than calculation led me to expect.  Speech has been possible not only to Paris but through Paris to Bruxelles, and even, with difficulty, through Paris to Marseilles, a distance of over 900 miles.  The wires between Paris and Marseilles are massive copper wires specially erected for telephone business between those important places.

4. Business Done.—­The charge for a conversation between London and Paris is 8 s. for three minutes’ complete use of the wire.  The demand for the wire is very considerable.  The average number of talks per day, exclusive of Sunday, is 86.  The maximum has been 108.  We have had as many as 19 per hour—­the average is 15 during the busy hours of the day.  As an instance of what can be done, 150 words per minute have been dictated in Paris and transcribed in London by shorthand writing.  Thus in three minutes 450 words were recorded, which at 8 s. cost five words for a penny.

5. Difficulties.—­The difficulties met with in long-distance speaking are several, and they may be divided into (a) those due to external disturbances and (b) those due to internal opposition.