[donee]

Hmm...,

The rain paths are different than I saw when I had tufts on my 2006 Prius. There was still a side ways component to the air velocity in the upper 1/3 of the A pillar area of the windshield.

The hump in the middle might be allowing the air on either side of the bump, on the windshield to squeeze into a lower pressure area formed behind the bump. Thus the inward velocity on the inboard side of the rain patterns. This air being inward, reduces pressure along the A-pillar, and avoids the spillage up high, but promotes it down low.

This seems to be backward to what Toyota did in the Gen III Prius, where they hollowed out the hood in the central area near the windshield. Interestingly, I took the turbulators I had on my Gen III in the upper 1/3 A-pillar area (similar to the Gen II) when I noticed the salt stains were straight back above and below the turbulator, and there were were no stains behind the turbulator. Indicating flow detachment. The windshield of the Gen III seems to be inset further in this area. And that seems to be helping the air go straight, and cascade smoothly when it goes over the edge of the body on either side of the windshield.

This does not mean what your doing is wrong, it just means the two circumstances are, most likely.

The tank averages are useless. You need need to do A/B/A coast down tests. The wind and temperature changes this time of year will completely skew the tank mileage averages....