Mike,
the differences in current are in order of 40 - 60%, that is significant.
The lower the band, the shorter the antenna, the bigger the effect, the more
important where the coil is. It will vary from antenna to antenna, depending on
the coil "shortening" factor. If the coil is closer to the feedpoint, the
current difference is lees, but efficiency suffers most. As you move coil up
the radiator, turns increase, current difference increases and effciency goes
up. If you replace (part of) coil with top loading, current differences
decrease (0 difference at 0 deg. long coil) and your efficiency goes up.
Efficiency or radiated power of loaded antenna is roughly proportional to the
area under the corresponding current curve of the remaining (straight)
radiator. Coil "eats" part of the radiator and its current carrying (radiating)
capabilities, this is why the current will be significantly different at the
ends of the coil. I hope this illustrates the situation?

As Cecil showed, modeling is not accounting for the effect and now that Roy is
on, we hope to sort things out and come up with ways to best implement the
phenomena in modeling programs. Right now, it appears that the best way to
approximate the effect is to use loading stubs of the same inductance as
intended coil.

Barry and Cecil agreed to cooperate on the article describing in detail (and in
civil manner :-) this subject and we hope that Roy will join us adding the
modeling aspect to it.

Yuri, K3BU/m