The driving motor is doing useful work (spinning itself and the "load" motor) and this takes power, that means a non zero phase current is flowing in the driving motor. And, as the driving motor cannot have a zero phase resistance, than means there is a voltage drop across the phases.
The "Load" motor is, in this case, effectively unconnected (assuming you were measuring the voltage with a high impedance multimeter!) and hence has zero phase current, and so no voltage drop.
So, the Back EMF measured from the second motor is of course lower than the voltage supplied to the first motor!
For example, lets say at speed X, with both motors being identical (and ignoring any unsymmetrical brush effects) the KE (the motors back emf vs rotational speed) of each motor generates a back emf of 10 volts. In the driving motor, there is a positive phase current, of say 1 amp, and if the winding impedance were 1 ohm, then 1 volt would be required to push that current through the phase windings, and so you'd have to apply 11 volts to the driving motor, yet get just 10V out of the load motor.