Imagine someone talking to you in a load bar. That person can, by shouting at you, make themself understood - but just barely. Although the signal is high, the noise (bar music, other conversations etc) is also high, and the quality of the communication is much reduced. That same person talking in a library could use a tone just barely above a whisper, and still be more clearly understood because the noise floor is so much lower in a library.

Many people don't realise that noise is actually the limiting factor in communications - if a signal has been attenuated, we can always just amplify it back up, but amplification amplifies the noise as well as the signal. Eventually, once the signal and the noise reach roughly equal amplitudes, amplification just gives you the noisy bar situation, you can't ever get back to the quiet library...

Bringing this back to the iPhone's antenna, we same to be in the library case, rather than the bar case. The signal that is being received when the antenna is being held is much quieter than what we see in other phones, but when we look at the only metric that really counts - bandwidth - we see that the iPhone is achieving consistently high transfer speeds, even when the antenna spacing has been bridged by someone's hand.

If I had to guess, based on the anandtech data, what we are seeing is an antenna that performs extremely well at reducing noise from the environment. Alternatively, the calibration of the sensor giving the signal strength readings in the 4G might not be terribly good. Either way, the independent tests done so far have all indicated that the iPhone 4G is outperforming every other phone out there at the moment, when looking at signal throughput.