412 Proceedings of Royal Society of Edinburgh, [sess. 
obtained, viz., 36° *2 Fahrenheit, and as this temperature appears 
to be practically the same for all depths exceeding, say, 2300 
fathoms, the conditions are exceptionally favourable for comparing 
the mean temperature given by thermometer with that obtained 
from electrical observations on the laid cable. 
Owing to the small and steady difference of earth potential 
between Bermuda and Turk’s Islands it was possible to determine 
the copper resistance of the laid cable with great accuracy, and 
the results so obtained were further checked by comparison with 
standard coils. 
The copper resistance at 75° of the different coils of core used in 
the manufacture of the cable was separately determined by the 
contractors and by ourselves with different instruments, and as 
very little difference was found between these results there is 
every reason to believe that the electrical values are accurate. 
The application of Matthiessen’s co-efficient for the resistance of 
copper at different temperatures, to the observed resistance of the 
laid cable, gave a mean temperature of 33° *3 Fahrenheit, while the 
mean observed temperature by thermometer was 36° *5 Fahrenheit. 
It appeared to us at first that this discrepancy must be due to an 
error in the thermometer values, but after reading Professor Tait’s 
Report on the Pressure Errors of the * Challenger ’ thermometers,* 
and having had the instruments used on the ‘ Britannia ’ verified 
at Kew, this did not seem possible. The only other explanation 
which occurred to us was that the temperature co-efficient of the 
copper used in the conductor was different to that hitherto em- 
ployed in calculation, and it was accordingly decided to investigate 
this point experimentally. 
A determination of the co-efficient was carried out with great 
care by Mr H. A. Taylor, who, through the courtesy of Mr W. 
Shuter, the Managing Director of the Telegraph Construction and 
Maintenance Co., was enabled to test samples of wire supplied by 
five of the leading manufacturers. The results obtained from these 
tests, which gave singularly even values for the different samples, 
showed that the temperature co-efficient for this class of copper was 
sensibly different from that obtained by Matthiessen, and the new 
co-efficient applied to the observed resistance of the laid cable 
* Narr. Chall. Exp., vol. ii. , Appendix A, 1882. 
