My experience has been that if you set the second above the first setting it actually increases the ink load on the paper but if you lower it below the first setting it does nothing.
Sounds odd. Normally in profiling one would set the test chart limit a little higher than the final target limit (10% is the rule of thumb ArgyllCMS uses by default), so that the measured data encompasses the final target gamut.
In my experience setting a final target above that used in the test chart results in poor accuracy - the device response has to be extrapolated, and extrapolation tends to be inaccurate.