DIRECT FEEDING SYSTEMS. 75 
which produce maximum loads seldom operate all over the 
system at once. 
With these data we can attack the feeder problem after 
deciding on the amount of copper to be put into the trol- 
ley wire and the value to be assigned to the track return. 
Step 5. How large ought the trolley wire to be? The 
answer to this question must be somewhat empirical, but 
we can get a line on it by considering the currents it hasto 
carry. Adopting the ladder system of Fig. 38 a very small 
trolley wire would answer. But we have seen that this 
arrangement is of little service in equalizing the voltage 
along the line, and hence it is better on the whole to use 
the system of Fig. 41 or some modification of it. Toavoid 
running an inconvenient number of feeders it is then de- 
sirable to install a trolley wire big enough to carry current 
for the service of a considerable distance. Referringnow to 
Plate 1, page 8, we see that allowing a drop of five per cent, 
i. e., twenty-five volts, in the trolley wire, all that should 
generally be tolerated at normal load, we can get reason- 
ably long distances between feeding points, say 3000 ft. or 
more, by using No. o or larger. No. oo is a standard size 
and gives rather better service than No. o in case of con- 
siderable load being bunched at one spot. Assuming this 
as the trolley wire, we may pass to the track return. 
The general principles of this have been very fully dis- 
cussed in Chap. II. ‘The ‘only thing needful here is to 
judge from the general conditions the value to be assigned 
to the conductivity of the track as compared with that of 
the overhead system. In the present case we are prob- 
ably dealing with sixty to seventy pound rails and the 
main line is double tracked. The bonding is, or should be 
made, good, and since the total service is not heavy the 
track conductivity is of the better class. It is probable 
therefore that raising the constant of equation 3, Chap. 
I, to 13 will fully take account of the return. Were the 
service even lighter or the rails continuous we might be 
justified in assuming 12, while with poor bonding and 
heavy traffic it might be necessary to assume 14.