358 Recent Advances in Telegraphy. [July, 
he wound it with two wires as similar as possible, but in 
opposite directions. It is unnecessary, then, to employ two 
different batteries. It is sufficient to split up the current 
from the signalling battery, sending one-half through one 
wire of the instrument into the line, and the other half 
through the other wire into the compensating circuit, which 
in this case must be made exactly equivalent to the line. 
This method of Frischen’s has been called the Differential 
Method. It was re-invented a few months later by Messrs. 
Siemens and Halske, of Berlin, and is therefore sometimes 
called the Frischen-Siemens method. 
Fig. 2 shows the arrangement of apparatus at either 
station. The receiving instrument is wound in opposite 
directions by two similar wires. The two adjacent ends of 
these two wires are joined to each other and to the lever of 
the key. The free ends are connected, one to the line, the 
other to the compensating circuit which — since it must be 
equivalent to the aCtual line-may now be called the 
Water Analogy. 
A l b 
“ artificial line. 59 It will be seen that on depressing the 
lever of the key to the battery-contaCt the current splits at 
the point where the wires of the instrument branch off, and, 
when the artificial line is electrically equal to the aCtual 
line, exactly one-half will flow through the wire (drawn full) 
into the aCtual line, while the other half will flow through 
the wire (drawn dotted) into the artificial line. These two 
halves will counteract each other’s effeCt on the needle, and 
the required eleCtric balance will be obtained. Another ad- 
vantage of this method over Gintl’s was that the line-circuit 
is never absolutely interrupted while keying goes on ; but it 
has this defeCt— that when the lever of the key rests on the 
earth-contaCt e the line is direCt “ to earth,’ 9 whereas when 
it passes from e to h the line is to earth through the artificial 
line. Such a variance in the earth-contaCts is attended by 
variance in the size of the received signals. 
A convenient means of elucidating the duplex principle, 
in the case of the differential method, is furnished us by a 
