NEC 2005, Article 450.4(A) has a confusing clause regarding autotransformer overcurrent protection:

An overcurrent device shall not be installed in series with the shunt winding (the winding common to both the input and the output circuits) of the autotransformer between Points A and B [Output taps] as shown in Figure 450.4.

What, exactly, do they mean by this? It sounds almost as though they're probihiting overcurrent protection on the output, but that makes no sense and this is directly contradicted by the table referenced earlier in the paragraph. Do people actually physically open up these transformers to break the windings so they can put a breaker in the middle, is that what they're prohibiting?