1. This division is a suitable one if δ≪ρ2−ρ1. As this is so in the practical cases of interest to us, we have used this division. It should be noted, however, that if the sign of the inequality is reversed it would be better to make the division on the disk z = δ. If this is done the analysis can be carried through in a way quite similar to that here reported.
2. One can be even more extreme and try the effect of using Ez = 1∕δ instead of (6). Although this is surely about the poorest approximation that might be used, the value of the main correction term σ2 is changed by less than ten percent. This justifies various assertions made earlier in the paper.