200 [June 19, 



maintained at that strength between the signals, the confusion of 

 interference would be avoided. He considered that, if the second or 

 earth-contact of each signal bore a fit proportion to the first contact, 

 this object might be effected ; and he considered that a succession of 

 very short pairs of contacts of a certain relative length, would main- 

 tain the current at the constant final strength during any pause sepa- 

 rating signals. He therefore prepared a paper band with openings cut 

 so as to make pairs of equal battery- and earth- contacts for dots, long 

 battery-contacts, followed by nearly equal earth-contacts, for a dash, 

 and a succession of pairs of very short contacts wherever a pause was 

 required, the battery contacts being rather the shorter of the two. 



The success of this plan was such that the signals were distinctly 

 recorded, not only by the galvanometer, but by a relay when the total 

 variations caused by the shortest signals were invisible on the galvano- 

 meter, i. e. even less than 1 per cent, of the maximum final current. 



7. Hence it was concluded that by the means adopted, or by 

 analogous means, " signals can be sent without confusion at any 

 speed which will allow the shortest signal used to cause a sensible 

 variation in the received current." 



These experiments were tried on dry cable coiled in iron tanks, and 

 might therefore not be applicable to extended and submerged cables. 



The author has, however, proved that the retardation and insula- 

 tion of an iron-covered cable are very little affected by the mere pre- 

 sence or absence of water ; and wherever the conclusions obtained 

 from the experiments agree with the deductions of theory, it is clear 

 that the experiments and theory confirm one another, and the con- 

 clusions may be safely applied to the practical case of a submerged 

 and extended cable ; for it is impossible to suppose that results due 

 only to an accidental arrangement of the cable should by chance 

 coincide with the deductions from a defective hypothesis. 



The experimental arrival-curves do not exactly agree with the 

 curve given by Professor W. Thomson (Proceedings of the Royal 

 Society, 1855. Phil. Mag. 1856). 



The experimental curve approaches its maximum much more slowly 

 than the mathematical curve, and continues to rise 1 or 2 per cent, 

 long after all effects from retardation as given by theory would cease. 



Some of this effect may be due to the mutual influence of the coils 



of the cable* ; but the greater part of the discrepancy is due to the 



* Vide paper read by Professor W. Thomson at the British Association, Aber- 



