Hi Brian,
The only thought I had to be honest, is how it's impossible to draw any conclusions since 97% of the noise will have been sampled out when the image was reduced for inline display. The only way you'll be able to see the difference is if you pixel-peep the originals, although having just said that, I'd eat my hat if the shot that was exposed +2EV didn't have noticeably less noise.
Photo 1 is a classic example of grossly under-exposed in terms of what's needed for best signal to noise ratio for a high ISO shot.