I have got the bottom of the mystery. At least I can sleep at night now.
A while back I ran small commentary on mysterious sagging lines on the side glass of 70s and 80s cars. This one was about the Mazda 1400 and this one showed the feature on the 1992 Mercedes S-class.
The nub of the issue is to do with how the designer wants the base of the glasshouse to be perceived. In the 1970s and 1980s car were more angular and upright than they are today. It was easy to run a line from the base of the windscreen around the A-pillar to the base of the sideglass or DLO as it is preposterously sometimes called.
Over the years cars’ profiles have been smoothed off. Furthermore the base of the windscreen has slowly raised relative to the base of the sideglass leading to a height differential that needed to be addressed. The options were to ignore the wraparound or to try to lead the eye around from windscreen to sideglass by some graphical means. A sagging line on the side glass was a subliminal way to do this and also, perhaps to give the impression by entasis that there was more horizontal curvature on the glass house.
2005 Peugeot 307 – the design breaks the side-to-side relationship.
Driven To Write 
Very good. Now explain to me the sagging valance of the mark 1 BMW 1 Series.
That’s a cause of some embarrassment to me. I asserted confidently in 1999 that you the sill or rocker panel had to be straight and a few years later BMW showed no such rule existed. It was an inspired critique of an unquestioned norm.
Perhaps you were right?
I’m with Chris, I think you were right about rocker panels being straight and not rocking-chair bowed like the BMW 1-er banana sill. I feel like I’m on the wrong side of the argument round here by harping on about contentious Bangled BMW styling like it’s 2005 but I have never liked that banana sill so at least I’m consistent.