After I put the meter back in the high leg is reading 320v to ground, 420v to phase B, 240 to C.
So the consensus is that this was caused by a polarity reversal which created a 120º phase shift instead of the 60º needed for a delta.
--I checked all the voltages before I began. 208v to ground 240-B, 240-C. 4 Wire. Only 1 taped red and 1 taped green. . Grounding rod comes in different raceway.
No, the B/C line are 120v to ground and 240v on phases. Only the high leg is testing off.
Perhaps there was an intermittent open connection between the transformer secondary windings that allowed the measurements of 208V, 240V, and 120V. Below is a possible explanation how this could happen even when the windings are reversed from the correct polarity.
As above, let A be the high leg with the two transformer windings tied together at phase C. Assume that the intermittent causes the windings to disconnect from eachother at C, but the C line output is still connected to the winding that provides the high leg A at its other end. This would allow the A-C untapped winding to "float" relative to the center-tapped B-C winding. If there is a load across B-C then that will establish a connection to the other side of the center-tapped winding at B, which is at 180º from C. Therefore both the A-B and A-C voltages would be 240V, and the A-N voltage would be 208V. The B-C voltage would be close to zero, but it was not clear from the above that this was measured.
Another possibility is that the transformer was wired the with correct polarity, and the intermittent open connection causes the same kind of inversion as above but with the disconnection on the other side of the tapped winding. This could explain having 320V A-N, but it wouldn't explain the 420V A-B and 240V A-C because that would require both sides of the tapped winding to be connected.