156 POWER DISTRIBUTION FOR ELECTRIC RAILROADS. 
ten per cent at full load, which is low, the net efficiency of 
the combination from the energy received by the motor to 
that delivered by the generator is only seventy-eight per 
cent. This loss of twenty-two per cent of the total energy 
in changing from alternating to continuous current is too 
considerable to be endured unless under very exceptional 
circumstances. It is quite out of the question if omne is 
considering the distribution of steam generated power for 
a long railway line or an extensive network, since the loss 
in power together with the cost of transmission would far 
overbalance any saving due to economy in generation, 
An alternative method is the use of a species of com- 
posite machine composed of alternating motor and contin- 
uous current dynamo assembled on the same base. As the 
two elements are rigidly in line, usually have one common 
bearing and are relieved from belt strain, their combined 
efficiency should be perhaps a couple of per cent better 
than would be indicated by their efficiencies taken separ- 
ately. Such composite machines are not used in this coun- 
try, but are not infrequently employed abroad, chiefly for 
lighting work. Fig. 89 shows a Siemens & Halske unit of 
this character for changing 2000 volt alternating to 150 volt 
continuous current. In wunits of 200 k. w. or So, such a ma- 
chine should show a full load efficiency of about eighty- 
eight per cent if properly designed. 
Sometimes windings for both kinds of current are put 
on a single armature core, but this device has little to rec- 
ommend it, 
For railway work by far the best method of obtaining 
continuous from polyphase current is by the use of the ap- 
paratus variously known as rotary transformer or rotary 
converter. ‘The principle of this machine can be readily 
seenl be reference to Fig. 72. If the armature here shown 
is put in rotation as an alternating motor by feeding alter- 
nating current into the collecting rings and bringing the 
machine into synchronism by any convenient means, there 
will evidently be flowing through the armature windings 
the same sort of current that would be generated if the