is alternating field really needed for NMR/ESR?

Consider a two level system - say a single spin-1/2 electron in a magnetic field. To make transition between the spin-split levels the usual advice is to apply an alternating transverse magnetic field at a frequency equal to the level splitting.

But what I don't understand is why the transverse field needs to be alternating at all. If I apply a static transverse field, I will see transitions to the other spin state as well. Why/how are these transitions different than the alternating field case? Why is one way 'better' than the other?

• You do realise that you need to apply a static magnetic field in the first place in any case, so that spin "up" and "down" have a different energy, don't you? – user154997 Oct 19 '17 at 15:28
• And if you aren't driving the system, why will it be switching between the split levels? – Jon Custer Oct 19 '17 at 16:10
• let me try to clarify: there is a field in z (take to be quantization axis). There is also a field in x. Start with spin in one state. Later it will be in a superposition of up and down which means there is a finite probability of a spin flip if measured. No ac field present. – user5419 Oct 19 '17 at 16:39
• just a guess: if you only apply a momentary bias along $x$ then the transition will be at random time and phase so the measured aggregate of all the transitions will be noise. – hyportnex Oct 19 '17 at 17:03